Gastro-Oesophageal Reflux Treatment for Prolonged

Gastro-oesophageal reﬂux treatment for prolonged non- speciﬁc cough in children and adults (Review)

Chang AB, Lasserson TJ, Gaffney J, Connor FL, Garske LA

This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2006, Issue 4 http://www.thecochranelibrary.com

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. TABLE OF CONTENTS HEADER...... 1 ABSTRACT ...... 1 PLAINLANGUAGESUMMARY ...... 2 BACKGROUND ...... 2 OBJECTIVES ...... 3 METHODS ...... 3 RESULTS...... 6 Figure1...... 7 Figure2...... 8 Figure3...... 9 Figure4...... 10 Figure5...... 10 Figure6...... 11 Figure7...... 11 Figure8...... 12 DISCUSSION ...... 13 Figure9...... 15 AUTHORS’CONCLUSIONS ...... 15 ACKNOWLEDGEMENTS ...... 16 REFERENCES ...... 16 CHARACTERISTICSOFSTUDIES ...... 21 DATAANDANALYSES...... 43 Analysis 1.1. Comparison 1 Thickened versus unthickened feeds (infants), Outcome 1 Subjects cured (of cough) at end of study...... 44 Analysis 1.2. Comparison 1 Thickened versus unthickened feeds (infants), Outcome 2 Cough frequency. . . . . 44 Analysis 2.1. Comparison 2 PPI versus placebo (> 18 years), Outcome 1 Clinical failures (still coughing at end of trial or reportingperiod)...... 45 Analysis 2.2. Comparison 2 PPI versus placebo (> 18 years), Outcome 2 Mean cough score at end of trial (1st arm crossover/parallelgrouptrials)...... 46 Analysis 2.3. Comparison 2 PPI versus placebo (> 18 years), Outcome 3 Change in cough scores (end-beginning of intervention - 1st arm crossover/parallel group trials)...... 47 Analysis 2.4. Comparison 2 PPI versus placebo (> 18 years), Outcome 4 Change in cough scores (crossover studies; standardisedscale)...... 48 Analysis 2.5. Comparison 2 PPI versus placebo (> 18 years), Outcome 5 Absolute cough scores (crossover studies, standardisedscale)...... 48 Analysis 2.6. Comparison 2 PPI versus placebo (> 18 years), Outcome 6 Change in cough score after 4 weeks treatment (1st arm cross over/parallel group trials)...... 49 Analysis 2.7. Comparison 2 PPI versus placebo (> 18 years), Outcome 7 Difference in cough scores at week 8 - week 4 (1st armcrossover/parallelgrouptrials)...... 50 APPENDICES ...... 50 WHAT’SNEW...... 51 HISTORY...... 51 CONTRIBUTIONSOFAUTHORS ...... 52 DECLARATIONSOFINTEREST ...... 52 SOURCESOFSUPPORT ...... 52 INDEXTERMS ...... 53

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) i Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. [Intervention Review] Gastro-oesophageal reflux treatment for prolonged non- specific cough in children and adults

Anne B Chang1, Toby J Lasserson2, Justin Gaffney3, Frances L Connor4, Luke A Garske5

1Queensland Children’s Respiratory Centre and Queensland Children’s Medical Research Institute, Royal Children’s Hospital, Brisbane and Menzies School of Health Research, CDU, Darwin, Brisbane, Australia. 2Community Health Sciences, St George’s, University of London, London, UK. 3Respiratory Medicine, Royal Children’s Hospital, Brisbane, Australia. 4Gastroenterology, Royal Children’s Hospital, Brisbane, Australia. 5Respiratory Medicine, Princess Alexandra Hospital, Brisbane, Australia

Contact address: Anne B Chang, Queensland Children’s Respiratory Centre and Queensland Children’s Medical Research Institute, Royal Children’s Hospital, Brisbane and Menzies School of Health Research, CDU, Darwin, Herston Road, Herston, Brisbane, Queensland, 4029, Australia. [email protected]. [email protected].

Editorial group: Cochrane Airways Group. Publication status and date: New search for studies and content updated (no change to conclusions), published in Issue 4, 2009. Review content assessed as up-to-date: 9 June 2009.

Citation: Chang AB, Lasserson TJ, Gaffney J, Connor FL, Garske LA. Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults. Cochrane Database of Systematic Reviews 2006, Issue 4. Art. No.: CD004823. DOI: 10.1002/14651858.CD004823.pub3.

ABSTRACT

Background

Gastroesophageal reﬂux disease (GORD) is said to be the causative factor in up to 41% of adults with chronic cough. However cough and GORD are common ailments and their co-existence by chance is high. Also cough can induce reﬂux episodes. Treatment for GORD includes conservative measures (diet manipulation), pharmaceutical therapy (motility or prokinetic agents, H2-antagonist and proton pump inhibitors (PPI)) and fundoplication.

Objectives

To evaluate the efﬁcacy of GORD treatment on chronic cough in children and adults with GORD and prolonged cough that is not related to an underlying respiratory disease i.e. non-speciﬁc chronic cough.

Search strategy

We searched the Cochrane Register of Controlled Trials (CENTRAL), the Cochrane Airways Group Specialised Register, MEDLINE and EMBASE databases, review articles and reference lists of relevant articles. The date of last search was 24th April 2009.

Selection criteria

All randomised controlled trials (RCTs) on GORD treatment for cough in children and adults without primary lung disease.

Data collection and analysis

Two review authors independently assessed trial quality and extracted. Study authors were contacted for further information.

Eighteen studies (5 paediatric, 13 adults) were included. None of the paediatric studies could be included in meta-analysis. In adults, analysis on use of H2 antagonist, motility agents and conservative treatment for GORD were not possible (from lack of data) and there were no controlled studies on fundoplication as an intervention. Nine adult studies comparing PPI (two to three months) to placebo were analysed for various outcomes in the meta-analysis. Enrolment of participants subjects for two studies were primarily from medical clinics and another eight studies were otolaryngology clinic patients or patients with laryngeal symptoms. Using “intention to treat”, pooled data from studies resulted in no significant difference between treatment and placebo in total resolution of cough, Odds Ratio 0.46 (95% confidence interval (CI) 0.19 to 1.15). Pooled data revealed no overall significant improvement in cough outcomes (end of trial or change in cough scores). Significant differences were only found in sensitivity analyses. A significant improvement in change of cough scores was found in end of intervention (two to three months) in those receiving PPI with a standardised mean difference of - 0.41 (95% CI -0.75 to -0.07) using generic inverse variance analysis on cross over trials. Two studies reported improvement in cough after five days to two weeks of treatment.

Authors’ conclusions

There is insufficient evidence to definitely conclude that GORD treatment with PPI is universally beneficial for cough associated with GORD in adults. The beneficial effect was only seen in sub-analysis. The optimal duration of such a trial of therapy to evaluate response could not be ascertained although two RCTs reported significant change by 2-weeks of therapy. Clinicians should be cognisant of a period (natural resolution with time) and placebo effect in studies that utilise cough as an outcome measure. Future paediatric and adult studies should be double blind, randomised controlled, parallel design, using treatments for at least two months, with validated subjective and objective cough outcomes and include ascertainment of time to respond as well as assessment of acid and/or non-acid reflux.

PLAIN LANGUAGE SUMMARY

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Cough in association with GORD is common in adults with chronic cough. The objective of this review was to evaluate the effectiveness of GORD treatment in children and adults with GORD and prolonged cough that is not related to an underlying respiratory disease i.e. non-specific chronic cough. Thirteen studies fulfilled predetermined criteria but only six could be used in various components of the meta-analysis including data obtained from trialists. Limited data on children prohibited any meta-analysis. In adults with cough and GORD, no significant difference was found in clinical cure using proton pump inhibitors (PPI) for cough and GORD. Using other outcomes, there was also no significant differences between PPI and placebo. This review also highlights the large placebo and time period effect of treatment for chronic cough. In adults GORD treatment with PPI for cough associated with GORD is inconsistent and its benefit variable. There was insufficient data to draw any conclusion from other therapies for cough associated with GORD. In children, the data is also inconclusive; thickened feeds had an inconsistent effect and no studies has examined PPI for cough and GORD.

(Chang 2001). Prolonged or chronic cough has been variously de- BACKGROUND fined as a cough which persists greater than three to eight weeks. Cough is the most common symptom presenting to general prac- titioners (Britt 2002). Worldwide, the desire to reduce the impact Gastroesophageal reflux (GOR) i.e. reflux of gastric contents into of the symptom of cough is reflected in the billions of dollars spent the oesophagus can be acid or non-acid (volume or alkaline reflux) on over the counter cough and cold medications (Morice 2002). and these occur physiologically, especially in the post-cibal state Non-specific cough has been defined as non-productive cough in (AMA 1996). When reflux is ’excessive’, GOR disease (GORD) is the absence of identifiable respiratory disease or known aetiology present. Symptoms attributed to GORD range from a variety of

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 2 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. gastrointestinal/ abdominal symptoms (abdominal pain, halitosis, mal reflux index; in adults the definition of abnormal pH metry water brash etc) to extra-oesophageal symptoms (cough, hoarse- consistent with GORD varies from a reflux index of greater than ness, laryngeal problems, ear disease, dental erosion etc) (AMA 3.9 to 7.2% (AMA 1996). 1996). Cohort studies in adults suggest that gastroesophageal A systematic review examining the efficacy of the various treatment reflux disease (GORD) related to acid, causes 21 to 41% of modalities for prolonged non-specific cough in adults and children chronic non-specific cough, including many with no gastrointesti- and GORD would help clarify the link (or otherwise) between nal symptoms of GORD (Irwin 1990). When successfully treated, cough and GORD. GORD associated cough is associated with a decrease in objective measurements of cough sensitivity (O’Connell 1994). In animals, potent acid instillation into the oesophagus causes airway neuro- genic inflammation (Daoui 2002). In adult humans similar experi- OBJECTIVES mentation causes cough in a temporal manner (within 70 seconds) To evaluate the effectiveness of GORD treatment in adults and (Ing 1994). Although a temporal relationship has been shown in children with GORD and prolonged cough not related to an un- the laboratory, cough is reported in some earlier studies to only derlying respiratory disease i.e. non-specific chronic cough. resolve after a mean time of 169 to 179 days following treatment for acid-GORD related cough (Irwin 2002a). It is thus not sur- prising that conflicting data have been reported and others have shown that acid GORD is associated with, but not the causative METHODS factor for, cough. Also, effective treatment of GORD that resolves gastrointestinal GORD symptoms may have no effect on objective pulmonary data (Ferrari 1995). Furthermore like asthma and Criteria for considering studies for this review GORD, both GORD and cough are common diseases, often coexist, and their association does not imply cause and effect (AMA 1996; Rudolph 2001; Rudolph 2003, Vakil 2006). Nevertheless, Types of studies several reviews/guidelines suggest a therapeutic trial of three to six All randomised controlled trials of any GORD treatment with months of GORD treatment for non-specific chronic dry cough cough as an outcome, where cough is not primarily related to an in both adults and children (Corrao 1996; Irwin 2002a; Kiljander underlying respiratory disorder (such as cystic fibrosis, asthma, 2003). chronic obstructive airway disease, suppurative lung disease etc) Current treatment for GORD include conservative anti-reflux or medication use (ACE inhibitor). measures (diet, positioning etc), pharmacological approaches [acid suppressing agents: histamine (H ) receptor antagonists, proton 2 2 Types of participants pump inhibitors (PPI) and prokinetic agents: domperidone, ery- thromycin, metoclopramide and previously cisapride] and surgi- Adults and/or children with chronic (three or more weeks) non- cal approaches (Nissen or Toupet fundoplication by laparoscopic specific cough (dry and non-productive cough without any other or open procedures) (Rudolph 2001). respiratory symptom, sign or systemic illness). Exclusion criteria: cough related to mycoplasma, pertussis and There are several definitions of GORD. Excessive distal oe- chlamydia, presence of underlying cardio-respiratory condition, sophageal acidification is the most common clinical scenario of current or recurrent wheeze (greater than two episodes), presence GORD and can be defined by the histological presence of reflux of other respiratory symptoms (productive cough, haemoptysis, oesophagitis (on biopsy) or distal oesophageal pH metry (24 hour dyspnoea), presence of other respiratory signs (clubbing, chest pH study) (AMA 1996). Distal pH metry is the most sensitive wall deformity, respiratory noises such as wheeze on auscultation method for defining acid GORD and is a standard test recom- and other adventitious sounds), presence of any sign of systemic mended for evaluating chronic cough (Chung 1996; Irwin 1998). illness (failure to thrive, aspiration, neurological or developmental In children, acid GORD is present if distal pH metry shows a abnormality), presence of lung function abnormality. reflux index (% time pH < 4) of > 10% in infants, and > 4% in children aged over 12 months as suggested by GORD guidelines by the European (Vandenplas 1993) and American paediatric gas- Types of interventions troenterology groups (Rudolph 2001). Dual channel (proximal All randomised controlled comparisons of therapies for GORD. and distal) oesophageal pH metry has also been used in assessing Trials only comparing two or more medications without a placebo GORD with arguable clinical use even in the assessment of pa- comparison group were not included. The following treatment tients at risk for upper airway complications of GORD (Rudolph regimes were evaluated: 2001). However there is variation in what constitutes an abnor- 1. Anti-reflux conservative measures

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 3 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. 2. H2 receptor antagonists (“gastro-oesophageal reflux” OR “gastroesophageal reflux” OR 3. Proton pump inhibitors (PPI) “gastro-esophageal reflux” OR “reflux” OR “ger” OR “gerd” OR 4. Surgical therapy “acid” OR “esophagus” OR “oesophagus”, all as [textword] or Trials that included the use of other medications or interventions [MeSH]) AND (“cough” as [textword] or [MeSH]) was included if all participants had equal access to such medica- For the full strategies please see Appendix 2; Appendix 3 and tions or interventions. Appendix 4. We identified trials from the following sources: 1. The Cochrane Controlled Trials Register (CENTRAL) which Types of outcome measures includes the Airways Collaborative Review Group Specialised Tri- Attempts were made to obtain data on at least one of the following als Register. outcome measures. 2. MEDLINE 1966 to 2009. Topic search strategy combined with the MEDLINE randomised controlled trial search filter as outlined in the Airways Group module. Primary outcomes 3. OLDMEDLINE 1951 to 1965. Topic search strategy combined with the Medline randomised controlled trial search filter Proportions of participants who were not cured or not substantially as outlined in the Airways Group module. improved at follow up (failure to cure). 4. EMBASE 1997 to 2009. Topic search strategy combined with the EMBASE randomised controlled trial search filter as outlined in the Airways Group module. Secondary outcomes 5. The list of references in relevant publications. 1. Proportions of participants who were not cured at follow 6. Written communication with the authors of trials included in up. the review when necessary. 2. Proportions of participants who not substantially improved Searches are current as of April 2009. at follow up. 3. Mean difference in cough indices (cough diary, cough frequency, cough scores). 4. Proportions experiencing adverse effects of the intervention, Data collection and analysis (e.g.. rash, surgical morbidity etc). 5. Proportions experiencing complications e.g.. requirement for medication change, repeat surgery etc. Selection of studies The proportions of participants who failed to improve on treat- From the title, abstract or descriptions, two review authors (ABC, ment and the mean clinical improvement were determined using LAG) independently reviewed literature searches to identify po- the following hierarchy of assessment measures (i.e. where two or tentially relevant trials for full review. Searches of bibliographies more assessment measures are reported in the same study, the out- and texts were conducted to identify additional studies. From the come measure that is listed first in the hierarchy was used). full text using specific criteria, the same two review authors inde- 1. Objective measurements of cough indices (cough pendently selected trials for inclusion. Agreement was measured frequency, cough receptor sensitivity). using kappa statistics. Disagreement was resolved by consensus. 2. Symptomatic (Quality of life, Likert scale, visual analogue scale, level of interference of cough, cough diary) - assessed by the patient (adult or child). Data extraction and management 3. Symptomatic (Quality of life, Likert scale, visual analogue scale, level of interference of cough, cough diary) - assessed by Trials that satisfied the inclusion criteria were reviewed and the fol- the parents/carers. lowing information recorded: study setting, year of study, source 4. Symptomatic (Likert scale, visual analogue scale, level of of funding, patient recruitment details (including number of el- interference of cough, cough diary) - assessed by clinicians. igible participants), inclusion and exclusion criteria, criteria used 5. Relevant airway markers consistent with inflammation. for GOR diagnosis, GOR symptoms, randomisation and allocation concealment method, numbers of participants randomised, blinding (masking) of participants, care providers and outcome assessors, dose and type of GORD therapy, duration of therapy, Search methods for identification of studies co-interventions, numbers of patients not followed up, reasons for The following topic search strategy was used to identify the withdrawals from study protocol (clinical, side-effects, refusal and relevant randomised controlled trials listed on the electronic other), details on side-effects of therapy, and whether intention- databases: to-treat analyses were possible. We extracted data on the outcomes

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 4 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. described previously. Further information was requested from the were included in the subsequent meta-analyses. In cross-over trials authors if required. (Eherer 2003; Kiljander 2000), when data was combined with parallel studies only data from the first arm was used as recommended (Elbourne 2002) There is methodological arguments of handling Assessment of risk of bias in included studies cross over data (Clarke 2003; Elbourne 2002) and we considered Two review authors independently performed quality assessment it was invalid to use second arm data given the known period effect on studies included in the review. Four components of quality of cough (i.e. cough tends to resolve with time (Chang 2001) and were assessed: carry-on effect (studies had short wash time of two weeks com- 1. Allocation concealment. Trials will be scored as; Grade A: pared to eight weeks of therapy in each arm). Adequate concealment, Grade B: Unclear, Grade C: Clearly For the dichotomous outcome variables of each individual study, inadequate concealment. (Grade A = high quality). relative and absolute risk reductions were calculated using a mod- 2. Blinding. Trials will be scored as: Grade A: Participant and ified intention-to-treat analysis. This analysis assumes that partic- care provider and outcome assessor blinded, Grade B: Outcome ipants not available for outcome assessment have not improved assessor blinded, Grade C: Unclear, Grade D: No blinding of (and probably represents a conservative estimate of effect). Num- outcome assessor (Grade A, B = high quality). bers needed to treat (NNT) was calculated from pooled data with 3. Reporting of participants by allocated group. Trials will be ’intention to treat’ used as the denominator and calculated using scored as: Grade A: The progress of all randomised participants the formula 1/risk difference (Clarke 2003). Event was defined as in each group described, Grade B: Unclear or no mention of successful treatment defined by absence of cough by end of treat- withdrawals or dropouts, Grade C: The progress of all ment period. randomised participants in each group clearly not described. Imputed correlation coefficient for calculation of standard devia- (Grade A = high quality). tion of change from baseline (Clarke 2003) was derived from other 4. Follow up. Trials will be scored as: Grade A: Outcomes studies in the review. The summary weighted risk ratio and 95% measured in > 90% (where withdrawals due to complications confidence interval (fixed-effect model) was calculated using the and side-effects are categorised as treatment failures), Grade B: inverse of the variance of each study result for weighting (RevMan Outcomes measured in 80 to 90%, Grade C: Unclear, Grade D: 2002). Data were pooled from similar studies, assuming cough Outcomes measured in < 80%. (Grade A = high quality) indices were normally distributed continuous variables. For cross- While only the allocation concealment quality assessment was dis- over studies, mean treatment differences were calculated from raw played in the meta-analysis figures, all assessments were included data, extracted or imputed and entered as fixed-effect generic in- in Characteristics of included studies. We measured Inter-review verse variance (GIV) outcome was used to give a weighted SD unit author reliability for the identification of high quality studies for difference and 95% confidence intervals (RevMan 2008) Hetero- each component using the Kappa statistic. geneity between the study results was tested to see if it reached Each study was also assessed using a 1 to 5 scale described by Jadad statistical significance using a chi-squared test. The 95% confi- et al (Jadad 1996) and summarised as follows: dence interval estimated using a random-effects model was in- 1. Was the study described as randomised? (1 = yes; 0 = no) cluded whenever there were concerns about statistical heterogene- 2. Was the study described as double blind? (1 = yes; 0 = no) ity. 3. Was there a description of withdrawals and dropouts? (1 = Gastrointestinal symptoms were not included in the forest plots yes; 0 = no) as it is not an objective of this review. 4. Was the method of randomisation clearly described and appropriate? (1 = yes; 0 = no) 5. Was the method of double blinding well described and appropriate? (1 = yes; 0 = no) Subgroup analysis and investigation of heterogeneity When authors of papers gave additional data, quality assessments Sub-group Analysis: and Jadad scores were based on additional data supplied. An a priori sub-group analysis was planned 1. Age: adults or children (aged < 18 years and < 12 months) 2. Definition of GORD used: acid GORD defined by pH metry or Data synthesis oesophageal biopsy or non acid/volume (alkaline) reflux or ’extra- An initial qualitative comparison of all the individually analysed oesophageal reflux’. studies examined whether pooling of results (meta-analysis) was 3. Intervention type: medical or surgical intervention reasonable. This took into account differences in study popula- Medical intervention further sub-grouped to: tions, inclusion/exclusion criteria, interventions, outcome assess- a) H2 antagonist; ment, and estimated effect size. The results from studies that met b) proton pump inhibitor (PPI); the inclusion criteria and reported any of the outcomes of interest c) conservative therapy.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 5 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Sensitivity analysis disease (hence did not have non-specific cough). FourTwo stud- We planned sensitivity analyses to assess the impact of the poten- ies (Orenstein 1992; Vanderhoof 2003; Chao 2007; Moukarzel tially important factors on the overall outcomes: 2007 ) reported on the use of specific anti-regurgitation formula 1. Sudy quality; milk (as opposed to thickened formula) in infants with GORD 2. Study size; that included cough as an outcome measure. However Moukarzel 3. Variation in the inclusion criteria; 2007 did not provide data specific to cough in the publication. 4. Differences in the medications used in the intervention and Orenstein and colleagues’ study described increased cough with comparison groups; thickened feeds as an intervention in a diverse group of infants 5. Differences in outcome measures; with cough and GORD (Orenstein 1992). Although most in the 6. Analysis using random effects model; group were otherwise well infants, some had primary respiratory 7. Analysis by “treatment received”; disease (hence did not have non-specific cough). Three studies 8. Analysis by “intention-to-treat”; and were supported (one in part) by the manufacturers of the milk 9. Analysis by study design-parallel and cross over studies utilised (Vanderhoof 2003; Chao 2007; Moukarzel 2007) and the (added after protocol written). other (Xinias 2003) was excluded because it was an open non-ran- We had planned to tested for publication bias using a funnel plot. domised (but controlled) trial. Outcome measures of the four studies on infants varied; in two he multi-centre studies (Vanderhoof 2003, Moukarzel 2007) cough was reported as part of a symptom complex (with gag or choke, etc) and although the authors provided further information, the data could not be used for analysis RESULTS in this review. In one study only a small number of infants had cough (9 of 81) (Chao 2007) displayed in outcome 1.1 (no significant difference between groups). Description of studies The single study in children (Dordal 1994) included children with asthma; its exclusion criteria were insufficiently defined to allow See: Characteristics of included studies; Characteristics of excluded classification of participants as having non-specific cough and it studies; Characteristics of studies awaiting classification. is unclear if the study was a randomised study. There were no controlled trials on the use of PPIs or surgery in infants or children. Results of the search The 2008 search resulted in 432 potential titles. Of these 10 papers were retrieved, 6 papers were excluded and 4 new studies Adults included. One of the included studies (Wo 2006) was identified In adults, thirteen studies were included; eleven were published from an excluded paper (Coron 2007) that was a generic review articles, one in abstract form (Kopec 2001) and one in a confer- of the subject. All the 4 studies (Chao 2007; Moukarzel 2007; ence report plus abstract (Ing 1997). Additional data were sought Pawar 2007; Wo 2006) were supported by the pharmaceutical in- from all but one author(s), six responded but only three (Eherer dustry involved in the product trialled in the study. This updated 2003; Kiljander 2000; Vaezi 2006) were able to provide additional Cochrane review now has a total of 18 included studies (14 from data that could be used for the meta-analysis. Although all studies previous version and 4 from current update). See Appendix 1 for included cough that was presumed associated with GOR, crite- details of results of previous searches. The 2009 search identified ria for entry into the studies varied. In six studies, participants 457 potential titles; 15 papers were retrieved but none fulfilled enrolled through the otolaryngology department had “laryngitis” inclusion criteria. Two studies were added to the excluded studies symptoms (Eherer 2003; Havas 1999; Noordzij 2001; Steward list. 2004; Wo 2006). One study based in medical clinics also primarily enrolled participants with chronic ’laryngitis’ (Vaezi 2006). In these studies where subjects were recruited from otolaryngology Included studies clinics, primary lung disease (such as asthma defined by hyper- responsiveness) as an exclusion criteria were not as stringent as they were in the studies enrolled through medical outpatients (Ing Paediatrics 1997; Jaspersen 1999; Kiljander 2000; Ours 1999). Mean age of There were five studies in children; four in infants and one in chil- participants reported in studies was 46 to 58 years (range 18 to 80 dren. Orenstein and colleagues’ study described increased cough years). Some studies had predominantly males (Eherer 2003; Ours with thickened feeds as an intervention in a diverse group of infants 1999) whereas others had predominantly females (Kiljander 2000; with cough and GORD (Orenstein 1992). Although most in the Steward 2004; Wo 2006). TwoOne study had similar proportions group were otherwise well infants, some had primary respiratory of males to females (Vaezi 2006; Pawar 2007). In all but one study

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 6 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. (Jaspersen 1999), presence of GOR (or extra-oesophageal reflux) evaluations during the trial (Vaezi 2006) but only one of these were confirmed objectively. However in Vaezi and colleagues study, studies provided ’during trial’ cough data in the paper (Noordzij only 29% of participants had pHmetry characteristics of GORD 2001). Objective cough monitoring was not used in any study. (Vaezi 2006). In most the diagnosis of GOR was made with pH- In all but one study, gastrointestinal or “extra-oesophageal” symp- metry, and some included dual channels. None of the studies used toms of GORD were also outcome measures. Adverse events were alkaline or volume reflux as an entry criteria although oesophageal specifically mentioned in six studies. manometry was also used in two studies. There were ten parallel studies and three crossover studies, both with wash out periods of two weeks. Ten studies compared PPI Risk of bias in included studies to placebo but varying doses and frequency were used. Studies involving participants with ’laryngitis’ (Eherer 2003; El Serag 2001; Jadad score of the studies varied from 1 to 5; six studies had scores Havas 1999; Noordzij 2001; Vaezi 2006; Steward 2004; Pawar of 5 (Eherer 2003; Ours 1999; Vaezi 2006; Vanderhoof 2003; 2007) generally used higher doses of PPIs (twice daily regime). Steward 2004; Wo 2006), three studies scored 4 (El Serag 2001; Three studies used omeprazole (Kiljander 2000; Noordzij 2001; Kiljander 2000; Noordzij 2001), three studies scored 3 points Ours 1999), one used esomeprazole (Vaezi 2006) and the other (Havas 1999; Ing 1997; Pawar 2007), three study scored 2 points three studies used a different PPI, pantoprazole (Eherer 2003; Wo (Orenstein 1992; Chao 2007; Moukarzel 2007), and three scored 2006), rabeprazole (Steward 2004; Pawar 2007 ) and lansoprazole one point (Dordal 1994; Jaspersen 1999; Kopec 2001). The qual- (Havas 1999; El Serag 2001). One study compared PPI to ran- ity score also varied: One study scored ’high quality’ in all four cat- itidine (Jaspersen 1999) and another was a two factorial design egories (Steward 2004). Five studies (Eherer 2003; El Serag 2001; using cisapride and diet intervention (Kopec 2001). In one study Ours 1999; Vaezi 2006; Wo 2006) scored ’high quality’ in three (Steward 2004) all subjects (i.e. both controls and intervention categories and four studies did have any high quality points for all groups) also received instructions to life style modification. There four categories. Agreement between the two main reviewers (AC and LG) for quality of studies was excellent; weighted kappa score was a single study on H2-antagonist versus placebo (Ing 1997), published only as a report on an oral presentation and abstract was 0.88 for Jadad score and 0.77 for quality assessment. format. Data from this study was presented graphically (provided In the studies in infants, randomisation and blinding were clearly by the author) and could not be used as points on the graph were described in two studies (Orenstein 1992; Vanderhoof 2003). unclear (number in the study did not equate to the number of Randomisation method and blinding were not mentioned in the points on the graph). There were no randomised controlled stud- other 2 studies in infants (Chao 2007; Moukarzel 2007) or in the ies on surgical intervention. study in children (Dordal 1994). In the adult studies, the randomisation process was clearly described in four but allocation conceal- Length of intervention in the study on H2-antagonist versus placebo was two weeks each (Ing 1997). In the studies using PPI, ment was unclear in all but one study (Steward 2004, Figure 1). length of intervention was two to three months (or 8 to 12 weeks) Method of blinding (i.e. appearance of placebo) was clearly de- and two study had prolonged follow up post trial (Ours 1999). scribed in five studies (Eherer 2003; Noordzij 2001; Ours 1999; Outcome measures for all studies were subjective cough scales of Vaezi 2006; Wo 2006). There were only four studies with data (in- varying types. Two studies had outcome assessments done mid cluding data that were sought from authors) that could be utilised way through trial (Havas 1999; Noordzij 2001), one had several as ’intention to treat’ for selected analysis (Eherer 2003; Kiljander 2000; Ours 1999; Vaezi 2006) without making any assumptions.

Figure 1. Methodological quality graph: review authors’ judgments about each methodological quality item presented as percentages across all included studies.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 7 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Effects of interventions study (Orenstein 1992) described significant increase in cough fre- The 18 studies included 823 randomised participants (372 paedi- quency post feeds when thickened feeds were given, Figure 2. The atrics,476 adults) with 506 (317 paediatrics, 412 adults) complet- multicentre study described decrease in percentage of feeds asso- ing the trials. Sixteen articles were published in English, one in ciated with cough/gag/choke episodes in infants given pre-thick- German (Jaspersen 1999) and another in French (Dordal 1994). ened milk (Vanderhoof 2003). Cough/gag/choke was grouped as In the updated searches of 2005, 2006 and 2007 three studies (El a secondary outcome and scored in a binary manner in the study Serag 2001; Vaezi 2006; Steward 2004) were found from recent of Vanderhoof and colleagues (Vanderhoof 2003). Additional in- review articles and the numbers above include these papers. Some formation was sought and provided but authors were unable to of the trialists responded to requests for data. Although entry cri- provide data on cough alone (cough was included in symptom teria were fulfilled in some participants, data on cough alone in complex with choke or gag or cough). In the Chao 2007 study, the these participants was not available on one study (El Serag 2001) number of children with cough was small (n=9 out of 81 infants but was available in the other two studies (Vaezi 2006; Steward who completed the study) and the difference between groups was 2004). not significant (outcome 1.1). In the Moukarzel 2007 study, although cough was an outcome measure, cough specific data were Paediatrics not presented in the paper. In the excluded study Xinias and col- There were insufficient data in infants and children to be dis- leagues found no effect of the anti-regurgitation formula milk on played in the MetaView graphs as trials in infants were too dis- cough (Xinias 2003). Dordal and colleagues described no signifi- similar to be included in a meta-analysis. All four studies in in- cant effect of cisapride or domperidone on cough presumably sec- fants found improvement in GORD symptoms referring to the GI ondary to GORD (Dordal 1994). No adverse events were reported system but data for effect on cough was inconsistent; the smaller in any study.

Figure 2. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.1 Clinical failures (still coughing at end of trial or reporting period).

Cough outcomes Adults

In the single study comparing PPI to H2-antagonist, 70% of those on PPI improved versus 30% of participants on H2-antagonist

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 8 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. (Jaspersen 1999). Data from the only study comparing H2-antag- Kiljander 2000; Ours 1999; Vaezi 2006) displayed in the forest onist to placebo was presented graphically and showed improve- plot. Pooled OR effect estimate was 0.46 and non significant (95% ment in cough scores by intervention in all participants and the CI 0.19 to 1.15, Figure 2). In Wo and colleagues study, the “% of effect was significant by two weeks (Ing 1997). subjects with adequate relief” was reported to be similar in both In the ten studies comparing PPI or cisapride to placebo, all but the PPI group and placebo group (40% and 42% respectively, p= two studies showed no difference between improvement of cough 0.89). As data specific for cough could not be obtained, this study scores in the active versus placebo arms (Eherer 2003; Havas was not included in the meta-analysis. 1999; Kopec 2001; Noordzij 2001; Ours 1999; Steward 2004; Vaezi 2006; Wo 2006). Only two studies described a significant difference in favour of PPI (Kiljander 2000; Pawar 2007). In the Mean cough score at end of trial (outcome 02.02) meta-analysis the outcomes are described below: In pooled analysis of four studies (Eherer 2003; Kiljander 2000; Noordzij 2001,Pawar 2007), the effect of active drug of borderline statistical significance (SMD -0.38, 95% CI -0.77, to 0, P = 0.05, Primary outcome: Failure to cure (outcome 02.01) Figure 3). A pooled analysis of data from crossover studies (Eherer This was the only outcome where “intention to treat” data could 2003; Kiljander 2000) was also not significant (SMD -0.29, 95% be utilised in all included studies for this outcome (Eherer 2003; CI -0.62 to 0.04, ).

Figure 3. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.2 Mean cough score at end of trial (1st arm crossover/parallel group trials).

Change in cough scores at end of intervention (parallel there was no significant heterogeneity, between group test for dif- group/1st arm crossover data) (outcome 02.03 to 02.05) ference between omeprazole and the other PPIs was non significant Pooled analysis for all six studies (Eherer 2003; Havas 1999; (SMD -0.45, 95% CI -1.13 to 0.23). This subgroup analysis was Kiljander 2000; Noordzij 2001;Steward 2004; Pawar 2007) was post hoc. Kiljander 2000 and Noordzij 2001, the two studies that statistically significant (SMD -0.39, 95% CI -0.77 to -0.08, Figure used omeprazole were also the studies that reported data at week 4), and there was a moderate level of heterogeneity (I square 12%). four and whilst the study drugs may have an effect over a longer The two studies (Kiljander 2000; Noordzij 2001) which utilised period than one month, the evidence from our analyses should omeprazole showed a significant benefit (SMD -0.71, 95% CI - be cautiously applied for several reasons. Firstly the sample sizes 1.29 to -0.41) and no significant heterogeneity. The studies that of the trials were small, and the sensitivity of the symptom scales utilised other PPIs (lansoprazole, rabeprazole or pantoprazole) is as yet unqualified. However, data from the two crossover trials (Havas 1999; Eherer 2003; Steward 2004;Pawar 2007) showed (Eherer 2003; Kiljander 2000) assessing mean change in symp- no significant benefit of PPI over placebo. Within each subgroup,

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 9 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. toms may confer some validity to the notion that there is an effect on symptom change. When these data were pooled using change from baseline scores, there was a significant difference of -0.41 standard deviation units (95% CI -0.75 to -0.07, Figure 5). The result with absolute scores was not statistically significant (Figure 6).

Figure 4. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.3 Change in cough scores (end-beginning of intervention - 1st arm crossover/parallel group trials).

Figure 5. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.4 Change in cough scores (crossover studies; standardised scale).

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 10 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Figure 6. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.5 Absolute cough scores (crossover studies, standardised scale).

Change in cough scores at week four of intervention (outcome 02.06) Two studies (Kiljander 2000; Noordzij 2001) could be utilised for assessment of time effect i.e. after four weeks of intervention. The change in cough score was in favour of PPI use but was not signiﬁcant (P = 0.09) with a standardised mean difference of -0.51 (95% CI -1.08 to 0.06, Figure 7).

Figure 7. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.6 Change in cough score after 4 weeks treatment (1st arm cross over/parallel group trials).

Change in cough score between week eight (end of intervention) and week four (mid-intervention) (outcome 02.07) Two studies (Kiljander 2000; Noordzij 2001) could be utilised for assessment of time effect from eight weeks (two months) to four weeks (one month) of intervention. There was no signiﬁcant difference between change in cough score between week eight and week four (SMD of -0.44 (95% CI -1.04 to 0.16), Figure 8.

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 11 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Figure 8. Forest plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.7 Difference in cough scores at week 8 - week 4 (1st arm cross over/parallel group trials).

Determination of time to respond and duration of treatment effect was limited. The two studies that reported on scores midway ences between PPI and placebo in hoarseness and throat clearing through trial (at one month (Noordzij 2001) and at six weeks but not in throat pain (Noordzij 2001). Another study (El Serag (Havas 1999)) found no difference in scores between midway 2001) described that a higher proportion of participants in the scores versus end of trial scores. In the meta-analysis, whereby raw PPI group (86%) had a complete response when compared to the data was obtained from Kiljander 2000 (data from Havas 1999 placebo group (40%) but no significant difference in laryngoscopy could not be included), no significant effect was found between assessment. One study described no difference between PPI and week eight and week four. In one crossover study, cough resolved placebo in total score but an improvement in cough score in favour while on omeprazole in two adults but recurred in the washout of PPI, yet the objective score in reflux (by video laryngoscopy) phase (Kiljander 2000). One study that reported time to response significantly favoured the placebo group (Pawar 2007). Wo and (Ours 1999) was also the only paper that had follow-up post trial colleagues described that at the 4 week follow up period post ces- (open non-RCT followed the double blind randomised placebo sation of treatment, those on PPI had a significantly higher recur- controlled parallel trial). The authors reported that cough totally rence of laryngeal symptoms then those on placebo (Wo 2006). resolved or showed a downward decline in cough scores in 5 to 14 When assessed by laryngoscopy, four studies described significant days in coughers who responded to open label PPI (Ours 1999). improvement in laryngoscopy-based scores but again the improve- None of the studies reported any significant adverse events to in- ment in those receiving PPI was similar to those on placebo (Eherer terventions and hence ’number needed to harm’ was not relevant 2003; Havas 1999; Steward 2004; Vaezi 2006). However, Noordzij in this review. Heterogeneity between studies included for meta- 2001 found no significant difference in laryngoscopy scores be- analysis was non-significant for all outcomes analysed. Funnel plot fore and after treatment in both arms. Pawar 2007 and Wo 2006 however looked asymmetrical but study numbers were small. As- described a significant difference between groups, both favouring sociation between level of risk and benefit was not possible. the placebo group.

Gastrointestinal symptoms of GORD Sensitivity analyses All studies with sufficient data provided reported significant im- Limiting analysis to studies with Jadad score of 5 did not changethe provement in GI symptoms over time but treatment arm was no primary outcome (proportions not cured) with an effect estimate different from placebo arm in five studies (Eherer 2003; Havas of 0.52 95% CI 0.20 to 1.35 (all-inclusive effect estimate 0.46, 1999; Noordzij 2001; Pawar 2007; Wo 2006 ). In two studies, 95% CI 0.19 to 1.15). It was not possible to perform this on improvement in GORD symptoms could not be determined from other outcomes. Varying inclusion criteria by removing studies lack of data or GORD symptoms were presumed absent (Ing 1997; of participants enrolled from otolaryngology clinics also did not Kopec 2001). change the primary outcome (effect estimate 0.17, 95% CI 0.02 to 1.73). Varying inclusion criteria by limiting studies to participants with ’laryngitis’ also did not change the primary outcome (effect Laryngeal symptoms and scores estimate 0.24, 95% CI 0.21 to 1.49). It was again not possible All eight studies that included this outcome measure reported to test this effect on other outcomes in the MetaView. Varying significant improvement in other laryngitis symptoms over time inclusion criteria by removing studies of participants enrolled from (Eherer 2003; El Serag 2001; Havas 1999; Noordzij 2001; Steward medical clinics did not alter non-significance of results (altered 2004; Vaezi 2006;Wo 2006 ). In six studies, the difference in to- ’change in cough score’ outcome from all-inclusive SMD -0.4, tal symptom improvement was similar in both the treatment arm 95% CI -0.86 to 0.06) to SMD -0.18, 95% CI -0.71 to 0.35). and placebo arms (Eherer 2003; Havas 1999; Steward 2004; Vaezi ’Mean cough score at end of trial’ did not significantly change when 2006; Pawar 2007; Wo 2006). One study found significant differ- medical clinic studies were omitted; SMD -0.34, 95% CI -0.97

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 12 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. to 0.28 and analysis was not possible for the primary outcome. 2000; Noordzij 2001) (both these studies also reported four weeks It was not possible to test for analysis for medication class type change) and in the GIV analysis on cross-over studies (Eherer or study size because of all studies were small and only studies on 2003; Kiljander 2000). Non significant changes were found in the PPI could be used in meta-analysis. Analysis by treatment received primary outcome measure of failure to cure, as well as in mean did not change the primary outcome i.e. proportions not cured; cough score at end of trial (borderline non-significant) and, when effect estimate 0.3 (95% CI 0.06 to 1.44). As analysis for other all studies were combined for change in cough score at end of outcomes were performed on “treatment received” (not possible trial analysis. While a possible contributor to this is the lack of on “intention to treat analysis”), this was not repeated for the power, another possible contributor to the inconsistent and small other three outcomes. There was no difference in results when treatment effect is that, as cough and GORD are both common random effects model was used in all outcome measures shown in symptoms, the presumed GORD related cough was not caused the MetaView. If however data from both arms of the cross-over by GORD. This explanation suggests that in a high proportion studies were treated as parallel studies, a significant difference was of cases of presumed GORD related cough, there is in fact an al- found between week eight and week four (SMD -0.76 (95% CI - ternative cause for the persistent cough. Indeed cough is the most 1.27 to -0.25) suggesting that the non significant effect is possibly common symptom presenting to doctors (Britt 2002). Both GOR related to a small sample size. and cough are common diseases, often co-exist and its association does not imply cause and effect (Field 1999). Indeed both symptoms co-existing merely by chance is high and cough can induce reflux episodes as described in asthma and GORD literature (Field DISCUSSION 1999; Zerbib 2002). However, our finding of an effect of therapy with PPI to improve cough does suggest that in a proportion of This systematic review of 18 studies (five in infants/children and cases of chronic cough with associated GORD, GORD is a con- thirteen in adults) has shown the lack of high-level evidence that tributing cause of cough. There is no other biologically plausible the treatment for GORD associated cough improves subjective explanation for the improvement in cough with PPI. cough in participants with non-specific cough universally. There was no effect in pooled analysis and the beneficial effect was seen Other possible influences of the results of this review include the only in the subgroup analysis. The OR for cough resolution pooled degree of acid inhibition achieved, the length of therapy, the out- from four studies was insignificant and if calculated, the number comes of cough measured, and the type of GORD (acid versus to treat to achieve cough resolution was high at 14.9 and the non-acid/volume reflux). The degree of acid inhibition achieved 95% confidence interval around the NNT ranged from a negative in the treatment is probably a small factor, if any, given that in the number to a positive one which hence includes possible harm. majority of studies resolution of GORD symptoms (of the gas- Hence the CI was not given (Altman 1998) and this emphasises trointestinal system) was achieved. However it is also possible that that treatment was not very different to placebo. This review also different degrees of acid suppression are required to control the highlights the large placebo and time period effect of treatment different manifestations of GORD. Subjective cough outcomes for chronic cough. were variable between studies and in all studies, the diary systems In contrast to the low effect of GORD treatment on cough found used were non validated systems. Subjective cough monitoring is in RCTs, in non-controlled trials (see ’Characteristics of excluded subjected to various influences (Chang 1998) and less reliable than studies’ table) the improvement rate of cough by surgical interven- objective cough monitoring in both adults (Hsu 1994) and chil- tion for GORD associated cough has been reported as high as 92% dren (Chang 2003; Chang 1998). In large RCTs, the influence of (Wright 2003) with cure rates as high as 81% (Brouwer 2003). For the outcome measures should be equal in both arms, but if one non-surgical intervention e.g. with PPI alone (Habermann 2002) medication is more likely to improve symptoms other than cough, or PPI with motility agents (Poe 2003), cough improvement rates cough specific symptom reporting may be unequally influenced. of 86 to 100% have been reported (Habermann 2002; Poe 2003). This is relevant as in the majority of included studies, resolution The large difference between the effect of treatment of GORD of other GORD symptoms was achieved, which may have influ- on cough seen in RCT and uncontrolled trails is likely related enced cough specific reporting. to the period effect and/or placebo effect also reported in other Some clinical heterogeneity was present in the participants of the treatments for cough (Chang 1999). The placebo effect of cough studies included as the majority but not all participants enrolled treatments has been reported to be as high as 85% (Eccles 2002). from otolaryngology clinics had cough. However using separate The high rate of placebo effect was specifically highlighted in two analysis, the direction of change in favour of PPI use was the of the included studies (Eherer 2003; Noordzij 2001). same in patients enrolled from medical or from otolaryngology The beneficial effects of PPI for cough and GORD was inconsis- clinics. Indeed excluding patients from otolaryngology clinics (or tent; significant only when cough outcomes were change in cough with laryngitis) did not alter the significance of the primary out- scores at end of trial of studies which used omeprazole (Kiljander come. There was no significant statistical heterogeneity in studies

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 13 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. between omeprazole and other PPIs, but this data is limited by PPIs are currently the most potent non surgical intervention for small sample size and different doses used by the different groups acid-GORD and no adverse events were reported in these studies. of trialists. The studies using omeprazole used either 40 mg daily The use of these medications as a trial of therapy thus probably has (Kiljander 2000) or bd (Noordzij 2001) while the dosage for lan- little risk. However an epidemiology study reported that adults on soprazole was 30 mg bd (Havas 1999), rabeprazole was 20 mg bd PPI or H2 antagonist have an increased risk of having community (Steward 2004) and pantoprazole was 40 mg bd (Eherer 2003). acquired pneumonia, adjusted relative risk of 1.89 (95% CI 1.36 to Other outcome analysis would not be possible without the studies 2.62) and 1.63 (95% CI 1.07 to 2.48) respectively (Laheij 2004). based in otolaryngology clinics, which would have resulted in non This risk translates to “approximately one case of pneumonia for significant changes in all the outcomes shown in the MetaView, every 100 years of patient exposure” (Gregor 2004). Another study with the exception of change in cough scores at the end of interven- described increased risk of hip fractures associated with PPI use; tion which was significant in the single medical clinic study with the adjusted odds ratio for hip fracture associated with more than data available. Another recent systematic review on laryngo-pha- 1 year of PPI therapy was 1.44 (95%CI 1.30-1.59) (Yang 2006). ryngeal reflux (LPR) (where cough is often a dominant symptom), Moreover, use of PPI has been reported to cause cough (Howaizi concluded that “high-dose proton pump inhibitor is no more ef- 2003) and the package insert for omeprazole includes cough as an fective than placebo in producing symptomatic improvement or adverse event in 1.1% of adults. In contrast to pharmaceutical in- resolution of laryngo-pharyngeal symptoms” (Gatta 2007). terventions, there were no studies on surgical intervention, which carries a small but significant rate of serious adverse events. Resolution of cough in response to effective treatment is the ideal outcome in these trials. However a clinically significant improve- Some authors suggest that cough in association with GORD re- ment, rather than total resolution of cough, may be relevant and lated to acid reflux can occur with normal pH metry and indices acceptable to patients with poor quality of life. The magnitude (i.e. require reflux index of 0%), and that cough can take a pro- of change in cough score that constitutes a clinically relevant im- longed time (a year) to settle post GORD intervention (Irwin provement is unknown. The NNT for clinical significant improve- 2002b). Such assertions are difficult to prove or disprove in the ment (as opposed to cough resolution) is unknown. The Cochrane context of the difficulties with using cough as the primary out- Airway Group’s policy is not to report on NNT (although arguably come measure in studies and the feasibility of the required studies. easily understood by clinicians) when the OR is not significant. The difficulties relating GORD to cough as causative (as opposed Under these circumstances, (Altman 1998) clearly states that when to an association) has been recently summarised (Eastburn 2007). CI for NNT should not be presented. However considerations of non acid reflux require further studies and perhaps multichannel intraluminal electrical impedance Data on length of therapy required to achieve a change in cough monitoring, said to be a more sensitive alternative technology for score are inconsistent. One study reported significant improve- evaluating all types of GORD (acid and non-acid) (Shay 2004) ment after two weeks of H2-antagonist in all participants (Ing may prove useful. 1997), which is a less potent acid suppressing agent than PPI (Rudolph 2001). Ours 1999 made specific reference to the re- Limitations of review sponse time of 5-14 days in those whose cough was relieved by PPI in the open label phase. However, in the meta-analysis limited The validity of this systematic review is hindered by the disparate to two studies where week four outcomes were possible (Kiljander nature of the interventions considered and the resulting small 2000; Noordzij 2001), a trend to improvement was seen at week number studies, with only selected availability of unpublished four and again at eight weeks but this was not significant. If data data. Given the distribution of effects for the primary outcome, from crossover trials were analysed as parallel trials (i.e. as first this may be a reason to suspect publication bias (Figure 9). Fur- arm data), there was a significant difference between week eight thermore although 18 studies were identified, data could only be and week four cough scores. While data on length of therapy is used for a subset of them. Our efforts to obtain numerical out- inconclusive, results from this review suggest that a trial of at least come data from a number of investigators have only been partially eight to nine weeks would be long enough to expect a significant successful. This review is also limited by a lack of validated scales improvement. This is also supported by a RCT comparing two and objective data on cough as well as a lack of allocation conceal- doses of lansoprazole (30 mg bd to 30 mg daily). In participants ment data and possibly by clinical heterogeneity of participants whose cough responded to lansoprazole, the response was seen by and medications. The review is primarily concerned with people four weeks and extended therapy (12 weeks) conferred no signif- who do not have primary lung disease but this as an exclusion icant additional benefit (Baldi 2006). There is no RCT data to factor could not be stringently applied in all studies. In all but four support recommendations of a six month therapy trial, in contrast studies (Steward 2004; Vaezi 2006; Pawar 2007; Wo 2006), par- to uncontrolled observations, that GORD associated cough takes ticipants were selected for gastro-intestinal symptoms or objective a mean of a five to six months to take effect (Corrao 1996; Irwin evidence of GORD and most but not all participants had cough. 2002a). GORD criteria also varied between studies which may influence

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 14 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. results. Most studies did not utilise GORD criteria specified by guidelines published by American and European Gastrointesti- nal Associations (AMA 1996; Rudolph 2001; Vandenplas 1993). There is also no data on non-acid reflux.

Figure 9. Funnel plot of comparison: 2 PPI versus placebo (> 18 years), outcome: 2.1 Clinical failures (still coughing at end of trial or reporting period).

small) risk of pneumonia in adults on acid suppressant therapy. If a therapeutic or empiric trial of PPI is undertaken for treatment AUTHORS’ CONCLUSIONS of cough, symptom improvement seems most likely to occur by Implications for practice two to eight weeks, although the optimal duration is uncertain. We would thus recommend that empiric trials not be abandoned Cough and gastrointestinal symptoms of GORD are common ail- before eight weeks until better evidence exist. However the period ments and presence of both symptoms by chance is high (in a co- effect (natural resolution with time) of cough is significant. Clin- hort of patients with chronic cough, the chance occurrence maybe icians should also be cognisant of the large influence of placebo as high as 25%) (Eastburn 2007). There is insufficient evidence intervention seen in studies that utilise cough as an outcome mea- to definitely conclude that adults that GORD treatment with PPI sure. It is probable that a proportion of adult patients with chronic for cough associated with GORD is beneficial. The beneficial ef- cough are subjected to longer term PPI, because of an initial ap- fect was only seen in sub-analysis and its effect small. Given the parent response due to this placebo effect, when their cough is in significant morbidity of chronic cough in many patients, a trial of fact not caused by their incidental GORD. There is insufficient therapy with PPI in adults with chronic cough and GORD maybe data to support (or refute) common recommendations of diet ma- considered but has to be balanced with the reported increased (but

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 15 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. nipulation or use of motility agents either as a single agent or in low-up time post trial to evaluate possible carry-on effect or recur- conjunction with PPIs in the management of GORD and cough. rence of cough post cessation of therapy. All three cross-over studies reported a significant carry-on effect (Eherer 2003; Ing 1997; In children, there is an absence of data on the utility of PPI for Kiljander 2003); cross-over designed studies should therefore not cough associated with GORD. Data on milk modification for in- be repeated. Objective measurement of reflux (acid and non-acid) fants and cough with GORD is also insufficient to make specific whilst on therapy would also be beneficial. Given the possible sig- recommendations. Until more evidence is available in the form of nificant harm of surgical intervention and the recommendation of well designed RCTs, other causes of cough should be considered surgical intervention if cough does not resolve with non-surgical in children with cough and GORD, prior to any consideration intervention for cough presumed associated with GORD symp- of empiric treatment with a prolonged course of GORD medica- toms (Irwin 1998), the need for a randomised controlled study tions/interventions. of surgical intervention is glaring. Respiratory illness and phar- macokinetics of medications (Sinaiko 2001) in children are suf- Implications for research ficiently different to that in adults to warrant separate studies on children using child appropriate valid outcomes (Chang 2003). Despite the widely advocated proposal that GORD associated cough is common and that prolonged treatment is required, there is paucity of RCT data on how effective GORD management is, in treating cough associated with GORD. Sufficiently pow- ACKNOWLEDGEMENTS ered RCTs examining time for improvement and resolution of cough as well as optimal duration of therapy using valid cough We thank Michael McKean and Chris Cates for their advice, sup- outcomes are required. These cough outcomes should include ob- portive role and comments to the protocol and review. We are also jective tools (e.g. ambulatory cough monitors) and validated sub- grateful to Elizabeth Arnold for performing the relevant searches jective cough instruments such as cough specific QoL instruments and obtaining the articles, and Toby Lasserson for translation of (Birring 2003; French 2002) and cough diaries (Chang 1998; Hsu German and French papers, Charlotta Pisinger for Czechoslo- 1994). Studies on prediction of response to treatment would also vakian translation and Gianni Ferrara for Italian translation. We be useful for clinical practice. The significant time period effect also thank Drs. Eherer, Ing, Kiljander, Kopec, Orenstein, El Serag, and placebo influences on cough as a symptom would render non- Omari, Steward and Vanderhoof for responding and/or provision controlled studies difficult to interpret. Design of future RCTs of additional data. For the 2007 update, we thank Susan Hansen should be parallel and placebo controlled and have sufficient fol- for performing the search as well as obtaining the relevant articles.

REFERENCES

References to studies included in this review ElSerag 2001 {published data only (unpublished sought but not used)} El Serag HB, Lee P, Buchner A, Inadomi JM, Gavin M, McCarthy DM. Lansoprazole treatment of patients with Chao 2007 {published and unpublished data} chronic idiopathic laryngitis: a placebo-controlled trial. Chao HC, Vandenplas Y. Comparison of the effect of a American Journal of Gastroenterology 2001;96(4):979–83. cornstarch thickened formula and strengthened regular formula on regurgitation, gastric emptying and weight gain Havas 1999 {published data only (unpublished sought but not used)} in infantile regurgitation. Diseases of the Esophagus 2007;20: Havas T, Huang S, Levy M, Abi-Hanna D, Truskett P, 155–60. Priestly J, et al.Posterior pharyngolaryngitis: Double- Dordal 1994 {published data only} blind randomised placebo-controlled trial of proton pump Dordal MT, Baltazar MA, Roca I, Marques L, Server inhibitor therapy. Australian Journal of Oto-Laryngology MT, Botoy J. Nocturnal spasmodic cough in the infant. 1999;3(3):243–6. Evolution after antireﬂux treatment [Toux spasmodique nocturne chez l’enfant. Evolution après traitement Ing 1997 {published and unpublished data} antireﬂux]. Allergergie et Immunology 1994;26(2):53–8. Ing A. Chronic cough. Respirology 1997;2(4):309–16.

Eherer 2003 {published and unpublished data} Jaspersen 1999 {published data only} Eherer AJ, Habermann W, Hammer HF,Kiesler K, Friedrich Jaspersen D, Diehl KL, Geyer P, Martens E. Omeprazole G, Krejs GJ. Effect of pantoprazole on the course of reflux- in reflux-associated chronic persistent cough [Omeprazol associated laryngitis: a placebo-controlled double-blind – (Antra MUPS) bei refluxassoziiertem chronisch– crossover study. Scandinavian Journal of Gastroenterology persistierenden Husten]. Endoskopie Heute 1999;12(2): 2003;38(5):462–7. 12–4.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 16 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Kiljander 2000 {published and unpublished data} gastroesophageal reflux. Clinical Pediatrics 2003;42(6): Kiljander TO, Salomaa ER, Hietanen EK, Terho EO. 483–95. Chronic cough and gastro-oesophageal reflux: a double- Wo 2006 {published data only} blind placebo-controlled study with omeprazole. European Wo JM, Koopman J, Harrell SP, Parker K, Winstead W, Respiratory Journal 2000;16(4):633–8. Lentsch E. Double-blind, placebo-controlled trial with Kopec 2001 {published and unpublished data} single-dose pantoprazole for laryngopharyngeal reflux.. Kopec SE, Irwin RS, French CL, Wilson MM, Bol S. American Journal of Gastroenterology 2006;101:1972–8. A double-blind randomized placebo-controlled trial References to studies excluded from this review comparing diet and/or cisapride. American Journal of Respiratory & Critical Care Medicine 2001;163(5 Suppl): Ahmad 2004 {published data only} A64. Ahmad I, Batch AJ. Acid reflux management: ENT Moukarzel 2007 {published data only} perspective. Journal of Laryngology & Otology 2004;118(1): Moukarzel AA, Abdelnour H, Akatcherian C. Effects of 25–30. a prethickened formula on esophageal pH and gastric Allen 1998 {published data only} emptying of infants with GER. Journal of Clinical Allen CJ, Anvari M. Gastro-oesophageal reflux related Gastroenterology 2007;41:823–9. cough and its response to laparoscopic fundoplication. Thorax 1998;53(11):963–8. Noordzij 2001 {published data only (unpublished sought but not used)} Noordzij JP, Khidr A, Evans BA, Desper E, Mittal RK, Allen 2002 {published data only} Reibel JF, et al.Evaluation of omeprazole in the treatment Allen CJ, Anvari M. Preoperative symptom evaluation and of reflux laryngitis: a prospective, placebo-controlled, esophageal acid infusion predict response to laparoscopic randomized, double-blind study. Laryngoscope 2001;111 Nissen fundoplication in gastroesophageal reflux patients (12):2147–51. who present with cough. Surgical Endoscopy 2002;16(7): 1037–41. Orenstein 1992 {published data only} Allen 2004 {published data only} Orenstein SR, Shalaby TM, Putnam PE. Thickened Allen CJ, Anvari M. Does laparoscopic fundoplication feedings as a cause of increased coughing when used as provide long-term control of gastroesophageal reflux related therapy for gastroesophageal reflux in infants. Journal of cough?. Surgical Endoscopy 2004;18(4):633–7. Pediatrics 1992;121(6):913–5. Baldi 2006 {published data only} Ours 1999 {published data only (unpublished sought but not used)} Baldi F, Cappiello R, Cavoli C, Ghersi S, Torresan F, Ours TM, Kavuru MS, Schilz RJ, Richter JE. A prospective Roda E. Proton pump inhibitor treatment of patients with evaluation of esophageal testing and a double-blind, gastroesophageal reflux-related chronic cough: a comparison randomized study of omeprazole in a diagnostic and between two different daily doses of lansoprazole. World therapeutic algorithm for chronic cough. American Journal Journal of Gastroenterology 2006;12(1):82–8. of Gastroenterology 1999;94(11):3131–8. Baldi 2006a {published data only} Pawar 2007 {published data only (unpublished sought but not used)} Baldi F, Cavoli C, Ghersi S, Mantovani L, Torresan F, Roda ∗ Pawar S, Lim HJ, Gill M, Smith TL, Merati A, Toohill RJ, E. Cost-effectiveness of different diagnostic strategies to Loehrl TA. Treatment of postnasal drip with proton pump assess gastro-oesophageal reflux disease in patients with inhibitors: a prospective, randomized, placebo-controlled unexplained chronic persistent cough in Italy. Digestive & study. American Journal of Rhinology 2007;21:695–701. Liver Disease 2006;38(7):452–8. Steward 2004 {published data only (unpublished sought but not used)} Belafsky 2008 {published data only} Steward DL, Wilson KM, Kelly DH, Patil MS, Belafsky PC, Rees CJ, Rodriguez K, Pryor JS, Katz PO. Schwartzbauer HR, Long JD, Welge JA. Proton pump Esophagopharyngeal reflux. Otolaryngology - Head and Neck inhibitor therapy for chronic laryngo-pharyngitis: a Surgery 2008;138:57–61. randomized placebo-control trial. Otolaryngology- Head & Brouwer 2003 {published data only} Neck Surgery 2004;131(4):342–50. Brouwer R, Kiroff GK. Improvement of respiratory Vaezi 2006 {published and unpublished data} symptoms following laparoscopic Nissen fundoplication. ∗ Vaezi MF, Richter JE, Stasney CR, Spiegel JR, Iannuzzi ANZ Journal of Surgery 2003;73(4):189–93. RA, Crawley JA, et al.Treatment of chronic posterior Chandra 2007 {published data only} laryngitis with esomeprazole. Laryngoscope 2006;116(2): Chandra KM, Harding SM. Therapy Insight: treatment 254–60. of gastroesophageal reflux in adults with chronic cough. Vanderhoof 2003 {published data only (unpublished sought but not Nature Clinical Practice Gastroenterology & Hepatology 2007; used)} 4:604–13. Vanderhoof JA, Moran JR, Harris CL, Merkel KL, Chen 2000 {published data only} Orenstein SR. Efficacy of a pre-thickened infant formula: a Chen RY, Thomas RJ. Results of laparoscopic fundoplication multicenter, double-blind, randomized, placebo-controlled where atypical symptoms coexist with oesophageal reflux. parallel group trial in 104 infants with symptomatic ANZ Journal of Surgery 2000;70(12):840–2.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 17 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Coron 2007 {published data only} Fraser 2000 {published data only} Coron E, Hatlebakk JG, Galmiche JP. Medical therapy Fraser AG, Morton RP, Gillibrand J. Presumed laryngo- of gastroesophageal reflux disease. Current Opinion in pharyngeal reflux: investigate or treat?. Journal of Gastroenterology 2007;23:434–9. Laryngology & Otology 2000;114(6):441–7.

Dalby-Payne 2003 {published data only} Gatta 2007 {published data only} Dalby-Payne JR, Morris AM, Craig JC. Meta-analysis of Gatta L, Vaira D, Sorrenti G, Zucchini S, Sama C, Vakil N. randomized controlled trials on the benefits and risks of Meta-analysis: The efficacy of proton pump inhibitors for using cisapride for the treatment of gastroesophageal reflux laryngeal symptoms attributed to gastro-oesophageal reflux in children. Journal of Gastroenterology & Hepatology 2003; disease. Alimentary Pharmacology & Therapeutics 2007;25 18(2):196–202. (4):385–92. Greason 2002 {published data only} DeMeester 1990 {published data only} Greason KL, Miller DL, Deschamps C, Allen MS, Nichols DeMeester TR, Bonavina L, Iascone C, Courtney JV, FC III, Trastek VF, et al.Effects of antireflux procedures on Skinner DB. Chronic respiratory symptoms and occult respiratory symptoms. Annals of Thoracic Surgery 2002;73 gastroesophageal reflux. A prospective clinical study and (2):381–5. results of surgical therapy. Annals of Surgery 1990;211(3): 337–45. Grill 1985 {published data only} Grill BB, Hillemeier AC, Semeraro LA, McCallum Dore 2007 {published data only} RW, Gryboski JD. Effects of domperidone therapy on Dore MP, Pedroni A, Pes GM, Maragkoudakis E, Tadeu V, symptoms and upper gastrointestinal motility in infants Pirina P, et al.Effect of antisecretory therapy on atypical with gastroesophageal reflux. Journal of Pediatrics 1985;106 symptoms in gastroesophageal reflux disease. Digestive (2):311–6. Diseases & Sciences 2007;52(2):463–8. Habermann 1999 {published data only} Duffy 2003 {published data only} Habermann W, Eherer A, Lindbichler F, Raith J, Friedrich Duffy JP,Maggard M, Hiyama DT, Atkinson JB, McFadden G. Ex juvantibus approach for chronic posterior laryngitis: DW, Ko CY, et al.Laparoscopic Nissen fundoplication results of short-term pantoprazole therapy. Journal of improves quality of life in patients with atypical symptoms Laryngology & Otology 1999;113(8):734–9. of gastroesophageal reflux. American Surgeon 2003;69(10): 833–8. Habermann 2002 {published data only} Habermann W, Kiesler K, Eherer A, Friedrich G. Short-term Ekstrom 2000 {published data only} therapeutic trial of proton pump inhibitors in suspected Ekstrom T, Johansson KE. Effects of anti-reflux surgery extraesophageal reflux. Journal of Voice 2002;16(3):425–32. on chronic cough and asthma in patients with gastro- oesophageal reflux disease. Respiratory Medicine 2000;94 Hui 2000 {published data only} (12):1166–70. Hui TT, Fass SM, Giurgiu DI, Lida A, Takagi S, Phillips EH. Gastroesophageal disease and nausea: does fundoplication El Hennawi 2004 {published data only} help or hurt?. Archives of Surgery 2000;135(5):545–9. El Hennawi DD, Iskander NM, Ibrahim IH, Serwah AH. Hunter 1996 {published data only} Persistent cough: prevalence of gastroesophageal reflux and Hunter JG, Trus TL, Branum GD, Waring JP, Wood WC. study of relevant laryngeal signs. Otolaryngology - Head & A physiologic approach to laparoscopic fundoplication for Neck Surgery 2004;131(5):767–72. gastroesophageal reflux disease. Annals of Surgery 1996;223 Eubanks 2001 {published data only} (6):673–85. Eubanks TR, Omelanczuk P, Hillel A, Maronian N, Pope Irwin 1993 {published data only} CE, Pellegrini CA. Pharyngeal pH measurements in patients Irwin RS, French CL, Curley FJ, Zawacki JK, Bennett FM. with respiratory symptoms before and during proton pump Chronic cough due to gastroesophageal reflux. Clinical, inhibitor therapy. American Journal of Surgery 2001;181(5): diagnostic, and pathogenetic aspects. Chest 1993;104(5): 466–70. 1511–7. Farrell 2001 {published data only} Irwin 2002 {published data only} Farrell TM, Richardson WS, Trus TL, Smith CD, Hunter Irwin RS, Zawacki JK, Wilson MM, French CT, Callery JG. Response of atypical symptoms of gastro-oesophageal MP. Chronic cough due to gastroesophageal reflux disease: reflux to antireflux surgery. British Journal of Surgery 2001; failure to resolve despite total/near-total elimination of 88(12):1649–52. esophageal acid. Chest 2002;121(4):1132–40. Fock 2008 {published data only} Issing 2004 {published data only} Fock KM, Talley NJ, Fass R, Goh KL, Katelaris P, Hunt Issing WJ, Karkos PD, Perreas K, Folwaczny C, Reichel R, et al.Asia-Pacific consensus on the management of O. Dual-probe 24-hour ambulatory pH monitoring for gastroesophageal reflux disease: Update. Journal of diagnosis of laryngopharyngeal reflux. Journal of Laryngology Gastroenterology and Hepatology 2008;23:8–22. & Otology 2004;118(11):845–8.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 18 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Katzka 1996 {published data only} Thoman 2002 {published data only} Katzka DA, Paoletti V, Leite L, Castell DO. Prolonged Thoman DS, Hui TT, Spyrou M, Phillips EH. Laparoscopic ambulatory pH monitoring in patients with persistent antireflux surgery and its effect on cough in patients with gastroesophageal reflux disease symptoms: testing while on gastroesophageal reflux disease. Journal of Gastrointestinal therapy identifies the need for more aggressive anti-reflux Surgery 2002;6(1):17–21. therapy. American Journal of Gastroenterology 1996;91(10): Tibbling 1993 {published data only} 2110–3. Tibbling L. Wrong-way swallowing as a possible cause of Leeder 2002 {published data only} bronchitis in patients with gastroesophageal reflux disease. Leeder PC, Watson DI, Jamieson GG. Laparoscopic Acta Otolaryngologia 1993;113(3):405–8. fundoplication for patients with symptoms but no objective Tibbling 1995 {published data only} evidence of gastroesophageal reflux. Diseases of the Esophagus Tibbling L, Gibellino FM, Johansson KE. Is mis-swallowing 2002;15(4):309–14. or smoking a cause of respiratory symptoms in patients with Monini 2006 {published data only} gastroesophageal reflux disease?. Dysphagia 1995;10(2): Monini S, Di Stadio A, Vestri A, Barbara M. Silent reflux: ex 113–6. juvantibus criteria for diagnosis and treatment of laryngeal van Zanten 2006 {published data only} disorders. Acta Oto-Laryngologica 2006;126(8):866–871. Van Zanten SV, Armstrong D, Chiba N, Flook N, White Murray 2006 {published data only} RJ, Chakraborty B, et al.Esomeprazole 40 mg once a day Murry T, Tabaee A, Owczarzak V, Aviv JE. Respiratory in patients with functional dyspepsia: The randomized, retraining therapy and management of laryngopharyngeal placebo-controlled ’ENTER’ trial. American Journal of reflux in the treatment of patients with cough and Gastroenterology 2006;101(9):2096–106. paradoxical vocal fold movement disorder. Annals of Waring 1995 {published data only} Otology, Rhinology & Laryngology 2006;115(10):754–8. Waring JP, Lacayo L, Hunter J, Katz E, Suwak B. Novitsky 2002 {published data only} Chronic cough and hoarseness in patients with severe Novitsky YW, Zawacki JK, Irwin RS, French CT, Hussey gastroesophageal reflux disease. Diagnosis and response to VM, Callery MP. Chronic cough due to gastroesophageal therapy. Digestive Diseases and Sciences 1995;40(5):1093–7. reflux disease: efficacy of antireflux surgery. Surgical Wo 1997 {published data only} Endoscopy 2002;16(4):567–71. Wo JM, Grist WJ, Gussack G, Delgaudio JM, Waring JP. Omari 2006 {published data only} Empiric trial of high-dose omeprazole in patients with Omari TI, Benninga MA, Sansom L, Butler RN, Dent posterior laryngitis: a prospective study. The American J, Davidson GP. Effect of baclofen on esophagogastric Journal of Gastroenterology 1997;92(12):2160–5. motility and gastroesophageal reflux in children with Wright 2003 {published data only} gastroesophageal reflux disease: a randomized controlled Wright RC, Rhodes KP. Improvement of laryngopharyngeal trial. Journal of Pediatrics 2006;149(4):468–74. reflux symptoms after laparoscopic Hill repair. American Oridate 2008 {published data only} Journal of Surgery 2003;185(5):455–61. Oridate N, Takeda H, Asaka M, Nishizawa N, Mesuda Y, Xinias 2003 {published data only} Mori M, Furuta Y, Fukuda S. Acid-suppression therapy Xinias I, Spiroglou K, Demertzidou V, Karatza E, Panteliadis offers varied laryngopharyngeal and esophageal symptom C. An anti-regurgitation milk formula in the management relief in laryngopharyngeal reflux patients. Digestive Diseases of infants with mild to moderate gastroesophageal reflux. & Sciences 2008;53:2033–8. Current Therapeutic Research, Clinical & Experimental 2003; Poe 2003 {published data only} 64(4):270–8. Poe RH, Kallay MC. Chronic cough and gastroesophageal Yang J 2006 {published data only} reflux disease: experience with specific therapy for diagnosis Yang JY, Lee HY, Kim NH, Kim YS. The effect of a proton- and treatment. Chest 2003;123(3):679–84. pump inhibitor in unexplained chronic cough patients. Songur 2008 {published data only} [Korean]. Tuberculosis & Respiratory Diseases 2006;61: Songur N, Songur Y, Cerci SS, Ozturk O, Sahin U, 137–42. Senol A, Yarktas MH. Gastroesophageal scintigraphy in the evaluation of adult patients with chronic cough References to studies awaiting assessment due to gastroesophageal reflux disease. Nuclear Medicine Communications 2008;29:1066–1072. Morice 2008 {published data only} Swoger 2006 {published data only} Morice AH, Donaldson JE, Fathi HH. The efficacy of Swoger J, Ponsky J, Hicks DM, Richter JE, Abelson dietary intervention in the treatment of reflux cough TI, Milstein C, et al.Surgical fundoplication in [Abstract]. American Thoracic Society International laryngopharyngeal reflux unresponsive to aggressive acid Conference. 2008; A897[#E121]. suppression: a controlled study. Clinical Gastroenterology & Additional references Hepatology 2006;4(4):433–41.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 19 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Altman 1998 probabilities, possibilities and limitations. Cough 2007;3:4. Altman DG. Confidence intervals for the number needed [DOI: 10.1186/1745-9974-3-4] to treat. BMJ 1998;317(7168):1309–12. Eccles 2002 AMA 1996 Eccles R. The powerful placebo in cough studies?. American Gastroenterological Associciation. American Pulmonary Pharmacology & Therapeutics 2002;15(3):303–8. Gastroenterological Association medical position statement: Elbourne 2002 guidelines on the use of esophageal pH recording. Elbourne DR, Altman DG, Higgins JP, Curtin F, Gastroenterology 1996;110(6):1981. Worthington HV, Vail A. Meta-analyses involving cross- Birring 2003 over trials: methodological issues. International Journal of Birring SS, Prudon B, Carr AJ, Singh SJ, Morgan MD, Epidemiology 2002;31(1):140–9. Pavord ID. Development of a symptom specific health Ferrari 1995 status measure for patients with chronic cough: Leicester Ferrari M, Olivieri M, Sembenini C, Benini L, Zuccali V, Cough Questionnaire (LCQ). Thorax 2003;58(4):339–43. Bardelli E, et al.Tussive effect of capsaicin in patients with Britt 2002 gastroesophageal reflux without cough. American Journal of Britt H, Miller GC, Knox S, Charles J, Valenti L, Henderson Respiratory & Critical Care Medicine 1995;151:557–61. J, et al.Bettering the Evaluation and Care of Health - A Field 1999 Study of General Practice Activity; 2002. Australian Institue Field SK, Sutherland LR. Gastroesophageal reflux and of Health and Welfare. Report no.: AIHW Cat. No. GEP- asthma: are they related?. Journal of Asthma 1999;36(8): 10. 631–44. Chang 1998 French 2002 Chang AB, Newman RG, Carlin J, Phelan PD, Robertson French CT, Irwin RS, Fletcher KE, Adams TM. Evaluation CF. Subjective scoring of cough in children: parent- of a cough-specific quality-of-life questionnaire. Chest completed vs child-completed diary cards vs an objective 2002;121(4):1123–31. method. European Respiratory Journal 1998;11:462–6. Gregor 2004 Chang 1999 Gregor JC. Acid suppression and pneumonia: a clinical Chang AB. State of the art: cough, cough receptors, and indication for rational prescribing. JAMA 2004;292(16): asthma in children. Pediatric Pulmonology 1999;28:59–70. 2012–3. Chang 2001 Howaizi 2003 Chang AB, Asher MI. A review of cough in children. Howaizi M, Delafosse C. Omeprazole-induced intractable Journal of Asthma 2001;38(4):299–309. cough. Annals of Pharmacotherapy 2003;37(11):1607–9. Chang 2003 Hsu 1994 Chang AB, Phelan PD, Robertson CF, Roberts RDG, Hsu JY, Stone RA, Logan-Sinclair RB, Worsdell M, Busst Sawyer SM. Relationship between measurements of cough CM, Chung KF. Coughing frequency in patients with severity. Archives of Disease in Childhood 2003;88:57–60. persistent cough: assessment using a 24 hour ambulatory recorder. European Respiratory Journal 1994;7:1246–53. Chung 1996 Ing 1994 Chung KF, Lalloo UG. Diagnosis and management of Ing AJ, Ngu MC, Breslin AB. Pathogenesis of chronic chronic persistent dry cough. Postgraduate Medicine 1996; persistent cough associated with gastroesophageal reflux. 72(852):594–8. American Journal of Respiratory & Critical Care Medicine Clarke 2003 1994;149(1):160–7. Clarke M, Oxman AD, editors. Formulating the problem. Irwin 1990 The Cochrane Reviewers’ Handbook 4.2 [updated November Irwin RS, Curley FJ, French CL. Chronic cough. The 2002]. The Cochrane Collaboration, 2003. spectrum and frequency of causes, key components of the Corrao 1996 diagnostic evaluation, and outcome of specific therapy. Corrao WM. Chronic persistent cough: diagnosis and American Review of Respiratory Disease 1990;141:640–7. treatment update. Pediatric Annals 1996;25(3):162–8. Irwin 1998 Daoui 2002 Irwin RS, Boulet LP, Cloutier MM, Fuller R, Gold PM, Daoui S, Agostino B, Gallelli L, Emonds X, Rossi F, Hoffstein V, et al.Managing cough as a defence mechanism Advenier C. Tachykinins and airway microvascular leakage and as a symptom. A consensus panel report of the induced by HCl intra-oesophageal instillation. European American College of Chest Physicians. Chest 1998;114(2 Respiratory Journal 2002;20:268–73. Suppl):133S–81S. Eastburn 2007 Irwin 2002a Eastburn MM, Katelaris PH, Chang AB. Defining the Irwin RS, Madison JM. Diagnosis and treatment of chronic relationship between gastroesophageal reflux and cough: cough due to gastro-esophageal reflux disease and postnasal

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 20 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. drip syndrome. Pulmonary Pharmacology & Therapeutics and treatment. American Journal of Medicine 2003;115 2002;15(3):261–6. (Suppl 3A):150S–156S. Irwin 2002b Shay 2004 Irwin RS, Zawacki JK, Wilson MM, French CT, Callery Shay S, Tutuian R, Sifrim D, Vela M, Wise J, Balaji N, et MP. Chronic cough due to gastroesophageal reflux disease: al.Twenty-four hour ambulatory simultaneous impedance failure to resolve despite total/near-total elimination of and pH monitoring: a multicenter report of normal esophageal acid. Chest 2002;121(4):1132–40. values from 60 healthy volunteers. American Journal of Kiljander 2003 Gastroenterology 2004;99(6):1037–43. Kiljander TO. The role of proton pump inhibitors in the management of gastroesophageal reflux disease-related Sinaiko 2001 asthma and chronic cough. American Journal of Medicine Sinaiko AR, Daniels SR. The use of short-acting nefedipine 2003;115(Suppl 3A):65S–71S. in children with hypertension: Another example of the Laheij 2004 need for comprehensive drug testing in children. Journal of Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Paediatrics 2001;139:7–9. Stricker BH, Jansen JB. Risk of community-acquired Vakil 2006 pneumonia and use of gastric acid-suppressive drugs. JAMA Vakil N, van Zanten SV, Kahrilas P, Dent J, Jones R. The 2004;292(16):1955–60. Montreal definition and classification of gastroesophageal O’Connell 1994 reflux disease: a global evidence-based consensus. American O’Connell F, Thomas VE, Pride NB, Fuller RW. Capsaicin Journal of Gastroenterology 2006;101(8):1900–20. cough sensitivity decreases with successful treatment of chronic cough. American Journal of Respiratory & Critical Vandenplas 1993 Care Medicine 1994;150:374–80. Vandenplas Y, Ashkenazi A, Belli D, Boige N, Bouquet RevMan 2008 J, Cadranel S, et al.A proposition for the diagnosis and The Nordic Cochrane Centre, The Cochrane Collaboration. treatment of gastro-oesophageal reflux disease in children: a Review Manager (RevMan). Version 5.. Copenhagen: The report from a working group on gastro-oesophageal reflux Nordic Cochrane Centre, The Cochrane Collaboration, disease. European Journal of Pediatrics 1993;152:704–11. 2008. Yang 2006 Rudolph 2001 Yang YX, Lewis JD, Epstein S, Metz DC. Long-term proton Rudolph CD, Mazur LJ, Liptak GS, Baker RD, Boyle pump inhibitor therapy and risk of hip fracture. JAMA JT, Colletti RB, et al.Guidelines for evaluation and 2006;296(Dec):2947–53. treatment of gastroesophageal reflux in infants and children: recommendations of the North American Society for Zerbib 2002 Pediatric Gastroenterology and Nutrition. Journal of Zerbib F, Guisset O, Lamouliatte H, Quinton A, Galmiche Pediatric Gastroenterology & Nutrition 2001;32(Suppl 2): JP, Tunon-de-Lara JM. Effects of bronchial obstruction on 1S–31S. lower esophageal sphincter motility and gastroesophageal Rudolph 2003 reflux in patients with asthma. American Journal of Rudolph CD. Supraesophageal complications of Respiratory & Critical Care Medicine 2002;166(9):1206–11. gastroesophageal reflux in children: challenges in diagnosis ∗ Indicates the major publication for the study

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 21 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. CHARACTERISTICSOFSTUDIES

Characteristics of included studies [ordered by study ID]

Chao 2007

Methods Single centre, double-blinded, randomised placebo-controlled trial that compared a commercial corn- starched milk anti-reﬂux (AR) formula to a regular 1.25% strength formula Infants given regular milk at baseline and scintigraphy performed. Diaries collected and repeat scintigraphy performed using the intervention formula Randomisation method not well described. Allocation by envelope Jadad score: 2, High quality score: B, C, A, B Study was ﬁnancially supported by a pharmaceutical industry

Participants 100 included infants, 81 completed the 2-month clinical follow-up. Intervention group (n=41): mean age-90.2 (SD 26.8) days, 21 males, 20 females. Placebo group (n=40): mean age-90.5 (SD 27.4) days, 21 males, 19 females. Only total of 9 children had cough as a symptom Inclusion criteria: Non breast-fed infants (age 2?4 months) presenting with frequent regurgitation/vomiting (> 3 times/day) Exclusion criteria: Infants with atopic symptoms such as eczema, watery rhinorrhoea or diarrhoea suspect- ing cow’s milk allergy, presence of mechanical obstruction such as infantile hypertrophic pyloric stenosis and malrotation (excluded with an upper gastrointestinal barium study)

Interventions A cornstarch thickened anti reﬂux formula compared to 1.25% strength regular formula for 2 months

Outcomes Diary: mean frequency of episodes of regurgitation, and/or vomiting, mean feeding volume tolerated as well as associated symptoms (irritability, cough, choking, crying) during the 3 days prior to a visit, and weight gain 90-min milk scintigraphy with 500 µCI technecium to quantify gastric emptying

Notes

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Envelope drawing system, no further information described

Methods Controlled parallel study comparing cisapride with domperidone and no treatment (as control). At baseline, patients were examined clinically. They underwent skin prick tests, spirometry, chest X-ray and gastro-oesophageal imaging (bolus liquid was administered and a computer generated image traced the path of the liquid through the oesophagus). Gastroesophageal imaging graded from 1-4. 1 = normal, 2 = partial retention of bolus (5 to 10 secs), 3 = retention of bolus (10 to 20 secs), 4 = prolonged retention of bolus (> 20 seconds). If gastroesophageal imaging of child >= 2, child eligible Dropouts: n = 10, 15.4% of those recruited. 4 dropouts from cisapride group (2 from adverse events, 2 lost), 2 from domperidone (lost) and 4 from no treatment group (lost) Randomisation not mentioned, blinding not described and allocation method not described

Participants 65 children (number screened not given) from outpatients with predominant nocturnal cough (with or without asthma) and GORD symptoms. 15 of the 55 participants who completed trial had cough without symptoms of asthma Median age of groups: cisapride- 6.7 years, SD 2.97 (n = 21, 15 males 6 females); domperidone - 7.05 years, SD 3.05 (n = 23, 10 males 13 females); control - 6.4 years, SD 2.62 (n = 11, 5 males 6 females) Inclusion criteria: Children with cough (with or without co-exiting asthma), predominantly nocturnal symptoms, reﬂux symptoms Exclusion: none described.

Interventions Cisapride (0.2mg/kg 20mins before each meal) with domperidone (0.2mg/kg 15mins before each meal) and no treatment (as control) for 12 weeks. All also received the following: Withhold food 2 hours before bedtime, reduce intake of acidic foods, lower size of meals, but increase frequency, raise height of the head of the bed by 10 cm and, sleeping on one’s side

Outcomes 1. Symptoms of cough 2. Gastro-imaging

Notes Data on children without asthma were not provided. No signiﬁcant difference found between the groups

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Methods Randomised double blind, crossover study comparing pantoprazole vs placebo in adults with ’laryngitis’ symptoms (includes cough) and GORD (deﬁned on pHmetry). Patients with inclusion criteria enrolled from outpatients clinic, telescopic video laryngo-stroboscopy performed and screened for exclusion criteria. Dual channel pHmetry performed and those with excessive reﬂux (> 4.5% time pH < 4) were eligible and randomised Randomisation by blocks of 4 performed by pharmaceutical supplier. Allocation method not described. Compliance monitoring not described Dropouts (30% of those randomised) were not described and were not included as treatment failures in paper. But analysis by ’intention to treat’ was possible with additional data provided by authors

Participants 21 adults (mean age 48 years, range 20-70, 16 males, 5 females) randomised from eligible 22 patients; (62 patients screened) with GORD (diagnosed on pHmetry) and symptoms of laryngitis (cough, nocturnal cough, sore throat, hoarseness, sore dysphonic attacks, globus sensation) attending otolaryngeal clinic in a tertiary institution (Karl-Franzens University, Austria). 12 of the 21 had chronic cough, 7 were initially commenced on placebo, and 5 on pantoprazole. 4 of those with cough (25%) did not complete trial; 2 in each arm Inclusion: hoarseness for > 2 months, laryngitis or other laryngeal symptoms which includes cough Exclusion: smokers, other causes of laryngitis, prior operations on laryngeal area, laryngeal malignancy, COPD

Interventions Placebo or pantoprazole 40mg bd for 3 months, 2 weeks wash out followed by pantoprazole or placebo for 3 months

Outcomes 1. Laryngeal symptoms (0-72) = sum of frequency of each symptom by intensity of symptom. 2. Oesophageal symptoms (0-48) = sum of frequency of each symptom by intensity of symptom. 3. Laryngeal scoring. All scored at 2 weeks after completion of treatment phase.

Notes Raw cough scores were scores provided by Dr. Eherer.

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

El Serag 2001

Methods Randomised double blind , parallel study comparing lansoprazole vs placebo in adults with ’laryngitis’ symptoms (includes cough) with or without GORD (deﬁned on pHmetry). Patients with inclusion criteria enrolled from otolaryngology outpatients clinic. Video laryngoscopy, dual channel pHmetry, gastroscopy (some also had oesophageal manometry) performed Allocation method not described. Compliance monitoring by pill counting Dropouts (n = 2, 9% of those randomised), one from each arm.

Participants 22 adults randomised from 27 patients screened, with laryngitis with or without GORD (diagnosed on pHmetry) and symptoms of laryngitis attending otolaryngeal clinic in a tertiary institution (Houston, Michigan). 14 (64%) of participants had chronic cough. Distal reﬂux deﬁned as DeMeester score > 14. 7. 21 males, 1 female in study, mean age 59 (SD 12) years in lansoprazole group, 65 (SD 12) years in

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 24 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. El Serag 2001 (Continued)

placebo group Inclusion: has (a) ’laryngitis’ symptoms as described in outcome measures, (b) posterior laryngitis as assessed by video laryngoscopy and (c) absence of concurrent infections or allergic causes of laryngitis Exclusion: aerodigestive malignancies, radiation therapy, or previous GI surgery

Interventions Lansoprazole 30mg bd or placebo bd for 3 months

Outcomes 1. Symptoms of hoarseness, frequent clearing of throat, dry cough, globus or persistent sore throat for > 3 weeks. 2. Laryngoscopy ﬁndings. (no scale used for either)

Notes Authors contacted for further information but no data were provided

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Havas 1999

Methods Randomised double blind, parallel study comparing lansoprazole vs placebo in adults with ’posterior pharyngolaryngitis’ symptoms (includes cough) with or without GORD (deﬁned on pHmetry). Patients with inclusion criteria enrolled from otolaryngology outpatients clinic. Video laryngoscopy, dual channel pHmetry, gastroscopy and oesophageal manometry performed Randomisation and allocation method not described. Compliance monitoring not described Dropouts (n = 5, 25% of those randomised) data on which group was not given and not included in paper’s analysis or described as treatment failures

Participants 20 adults randomised from 100 patients screened, with laryngopharyngeal reflux with or without GORD (diagnosed on pHmetry) and symptoms of laryngitis attending otolaryngeal clinic in a tertiary institution (Prince of Wales Hospital, Sydney). Distal reflux defined as pH< 4 for > 4% of time, proximal = pH fall of >= 3 within nadir of =< 5 with oesophageal acidification. Of the 15 (7 males, 8 females, mean age 52. 9 years) who completed the trial, 8 received lansoprazole and 7 placebo Inclusion: has ’posterior pharyngolaryngitis’ assessed by video laryngoscopy and symptoms (described in outcome measures) Exclusion: severe neurological disorders, chronic airflow limitation, pre-existing anti-secretory medications, severe oesophagitis seen at endoscopy, professional voice users (singers)

Interventions Lansoprazole 30mg bd or placebo bd for 12 weeks

Outcomes 1. Symptom scores (addition of scores for severity [0-3] to scores for frequency [0-4]) for each of 4 symptoms of laryngitis (chronic cough, hoarseness, throat clearing, sore throat pain). 2. Reﬂux symptoms scored as above for dysphagia, retrosternal burning pain, acid regurgitation and odynophagia. 3. Laryngoscopy ﬁndings (0-4) 4. Adverse events. Scores obtained at 6 and 12 weeks and data on cough alone was provided in paper

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 25 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Havas 1999 (Continued)

Notes Authors contacted for further information but no data were provided

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Ing 1997

Methods Double blind cross over study comparing ranitidine with placebo. Patients with chronic persistent cough (> 2 months) with unknown aetiology after a standard diagnostic evaluation were evaluated. They underwent pHmetry and histamine airway hyper responsiveness. pHmetry was repeated in 7 participants, 4-6 weeks after the ranitidine treatment period Dropouts (n = 1, 4% of those recruited).

Participants 25 patients randomised (number screened not given) from medical outpatients of a tertiary hospital. All patients had PD20 to histamine > 8 umols. 13 participants had placebo medication in the first treatment period, and the 11 patients who had ranitidine in the first treatment period (1 dropout described) Mean age was 46.8 (SEM 3.3) yrs, 11 male, 14 female. Inclusion: chronic cough and significant gastro-oesophageal reflux (presence of > 12 reflux episodes per 24 hours and time pH < 4 was > 0.7%) Exclusion: patients with bronchial asthma, chronic bronchitis, allergic rhinitis, sinusitis, post nasal drip, other respiratory and cardiac diseases, acute respiratory infections occurring within 2 months of the study, smokers, or using theophylline or angiotensin converting enzyme inhibitors

Interventions Placebo or ranitidine 150mg bd for 2 weeks, 2 weeks wash out followed by ranitidine or placebo for 2 weeks

Outcomes 1. Cough scores (1-4) on a twice daily on diary cards recorded basis 2. Presence of symptoms of GORD.

Notes Study not published other than in abstract and conference report

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Methods Randomised, parallel study comparison of omeprazole versus ranitidine. Method of allocation not described. Study was not double blinded Compliance monitoring not mentioned. No withdrawals reported.

Participants 20 adults from outpatients department, with chronic cough, with otherwise normal lung function and negative clinical and radiological findings. 10 participants in omeprazole group (5 males, 5 females, mean age 49.4 years SD 16.3) and 10 in ranitidine group (6 males, 5 females, mean age 46.5 SD 19.7) Inclusion: persistent cough for 8 weeks, negative findings from physical examination, negative radiological findings, with either symptoms of GORD or positive investigation of GORD (diagnosed by gastroscopy) , and normal lung function Exclusion: Diagnosis of chronic asthma, COPD, chronic bronchitis, other oesophageal disease, ACE inhibitor therapy, ENT abnormalities, or smokers

Interventions Omeprazole 10mg/day or ranitidine 300mg/day for 8 weeks.

Outcomes 1. Cough score (range 0-4) 2. Oesophagitis score (range 0-4) 3. Adverse events. Outcomes recorded for 8 weeks.

Notes

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Kiljander 2000

Methods Randomised double-blind, crossover study comparing omeprazole vs placebo in adults with chronic cough and GORD (deﬁned on pHmetry). Patients with chronic cough enrolled from pulmonary outpatients clinic were screened for exclusion criteria. Patients completed one-week observation by diary cards, pH- metry then performed and those with excessive reﬂux were randomised Randomisation and allocation method not described. Compliance monitored by pill counts after treatment period. 8 did not complete trial (27% of those randomised) were not further described and were not included as treatment failures. Analysis by intention to treat for primary outcome only

Participants 29 adults (median age 49, IQR 20-74, 10 males, 19 females) randomised from 48 patients screened, with GORD (diagnosed on pHmetry) and chronic (>= 2months) cough attending pulmonary clinic in a tertiary institution (Turku University, Finland). 12 initially randomised to placebo and 9 to omeprazole Inclusion: chronic persistent cough (>= 2 months) Exclusion: Abnormal chest or sinus radiology, positive methacholine test, rhinitis, nasal mucosa appearance of cobblestone or muco-purulent secretions, smokers, asthma, chronic bronchitis, use of angiotensin converting enzyme inhibitor

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 27 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Kiljander 2000 (Continued)

Interventions Omeprazole 40mg per day or placebo for 8 weeks, 2 weeks washout followed by omeprazole or placebo

Outcomes 1. Weekly daytime cough and night time cough (cough disturbing sleep) scores = sum of daily 4 point (0 to 3) scale of symptoms (both has range of 0-21) 2. Weekly gastric symptoms (heartburn, regurgitation, chest pain) = sum of daily 4 point (0 to 3) scale of symptoms (range 0-63). Average of last 3 weeks at end of each treatment period (8 weeks)

Notes Cough scores obtained from primary author.

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Kopec 2001

Methods Partial double-blind parallel randomised placebo controlled trial comparing diet and/or cisapride (2 factorial design). Randomisation of the diet was not blinded, randomisation of cisapride was double- blinded After standard diagnostic work-up for chronic cough, participants who met criteria (authors assume that this predicts patients have an approximate 92% chance of having chronic cough due to GERD) were randomised and pHmetry performed Randomisation and allocation method not described. Dropouts (n = 2, 9.5% of those randomised) were not described and were not included as treatment failures. Analysis by intention to treat not possible Compliance monitoring not described.

Participants 21 adults (mean age 53.5 years, SD 12.9) with chronic cough and GORD randomised from outpatients clinic. (number of eligible patients not given). GORD diagnosed on pHmetry or on barium meal Inclusion: Adults with chronic cough (> 8 weeks) and presumably GORD related cough Exclusion: age < 18 years, pregnancy, known contraindications to receiving cisapride, such as known allergy to the medication, or prolonged QT interval on EKG or concurrent use of medications that might interact with cisapride to place the patients at risk of a potentially life-threatening cardiovascular complication

Interventions 2 factorial design using usual care diet or anti-reﬂux diet and cisapride 10mg qid or placebo for 4 months trial ie 4 arm study

Outcomes Cough scores measured by visual analogue score (VAS) from 0 to 100

Notes No difference found in VAS between groups Further data provided by author was insufﬁcient for inclusion into meta-analysis

Risk of bias

Item Authors’ judgement Description

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 28 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Kopec 2001 (Continued)

Allocation concealment? Unclear Information not available

Moukarzel 2007

Methods Single centre, double-blinded, randomised placebo-controlled trial that compared a commercial pre- thickened anti-reflux (AR) formula to a regular formula Medical history was obtained and complete physical examination was performed at entry. All infants underwent a 24-hour oesophageal pH monitoring while receiving alternating normal and anti-reflux milk. Weekly monitoring of episodes of regurgitation, vomiting, coughing, crying, and stool frequency and consistency. The volume of each feeding during a 24 hours period once weekly was also recorded Randomisation and allocation method not well described. Jadad score: 2, High quality score: B, C, A, B Study was financially supported by a pharmaceutical industry

Participants 74 included infants, 60 completed the 1-month clinical follow-up. Mean age of group was 3.24 (SD 1. 28) months, 40 males, 34 females. Intervention group n=28, Placebo group n=32. Mean age of infants within each group was not described Inclusion criteria: Non breast-fed infants (aged <6 months) with GER (determined using Orenstein criteria) Exclusion criteria: Breast-fed and premature infants, infants with history of wheezing, aspiration pneumonia, apnoea, failure to thrive, anaemia, bleeding, laryngitis, and apparent life threatening events. Infants already receiving AR or medications that could affect the motility of the gastrointestinal tract. Parents who subjected infants to overfeeding, dilution errors, and inadequate feeding technique

Interventions A pre-thickened anti reﬂux formula compared to regular formula for 1 month

Outcomes Diary: episodes of regurgitation, vomiting, coughing, crying, and stool frequency and consistency Data relating to pH monitoring (reﬂux index, oesophageal clearance, etc), electrogastrography (a cutaneous recording of gastric myoelectrical activity

Notes Wrote to authors 31st May 2008

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Methods Randomised double blind, parallel study comparing omeprazole vs placebo in adults with ’laryngitis’ symptoms (includes cough) and GORD (deﬁned on pHmetry) Patients with inclusion criteria enrolled from outpatients clinic. Video-laryngo-stroboscopy performed and screened for exclusion criteria. Dual channel pHmetry then performed and those with reﬂux were eligible and randomised Randomisation and allocation method not described. Compliance monitoring not described. Dropouts (6.7% of those randomised) were included in analysis but not described as treatment failures. Analysis by intention to treat not possible

Participants 30 adults randomised from 53 patients screened) with GORD/laryngopharyngeal reflux (diagnosed on pHmetry) and symptoms of laryngitis attending otolaryngeal clinic in a tertiary institution (University of Virginia). 15 participants in omeprazole group (8 males, 7 females, mean age 51.7 years) and 15 in placebo group (8 males, 7 females, mean age 45.3). 2 (6.7 %) did not complete trial, one from placebo group and the other unknown Inclusion: one or more symptoms of laryngitis for > 3 months ie symptoms of chronic cough, hoarseness, excessive phlegm, throat clearing, throat pain, lump in throat and acid reflux (> 4 episodes proximal pH < 4 or 3 point drop in pH with simultaneous drop in distal pH of < 4) Exclusion: viral or bacterial laryngitis (undefined), benign vocal fold lesions, occupational exposures causing laryngitis, history of seasonal allergies or laryngeal malignancy

Interventions Omeprazole 40mg bd or placebo bd for 2 months.

Outcomes 1. Symptom scores (multiplication of severity [0-100] of symptom by frequency [number of days over last 2 weeks] for each of 6 symptoms of laryngitis (chronic cough, hoarseness, excessive phlegm, throat clearing, throat pain, lump in throat) and 3 symptoms of GOR (dysphagia, odynophagia and heartburn 2. Adverse events. Scores obtained at 1 and 2 months and data on cough alone was provided in paper

Notes Authors contacted for further information but no data were provided

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Orenstein 1992

Methods Randomised single blind, cross over study comparing thickened and unthickened feeds on cough associated with feeds, related to GORD diagnosed by pHmetry, barium swallow or oesophageal biopsy. Assessor was blinded but not care-giver/nurse who fed infants Randomisation well described (lottery) but allocation method not described No dropouts mentioned. Uncertain if analysis was by intention to treat

Participants 25 infants randomised (number screened not given) from outpatients of a tertiary gastroenterology centre (Children’s Hospital of Pittsburg) referred for GORD Median post natal age 7.5 weeks (range 2-26), corrected age 7 weeks (-6 to 26), gender not given. Symptoms

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 30 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Orenstein 1992 (Continued)

other than cough were irritability in 72%, regurgitation in 60%, apnoea in 28%, history of pneumonia in 4% Inclusion criteria: not deﬁned. Exclusion: concurrent acute infectious respiratory illness, use of prokinetic or acid reducing medications

Interventions Thickened (with rice cereal) and unthickened feeds.

Outcomes Cough frequency during feeds and post prandial per hour.

Notes Results showed that cough frequency when on thickened feeds (3.9) was signiﬁcantly higher than when on unthickened feeds (2.2); P = 0.006

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Ours 1999

Methods Randomised double blind, parallel comparison of omeprazole vs placebo in adults with chronic cough and GORD (deﬁned on pHmetry). Potential participants screened for exclusion criteria. Patients completed 2 weeks observation by diary cards. Those with score of at least 3 were treated with antihistamine and nasal corticosteroids. If persistent cough remained (score >= 2) oesophageal manometry and pHmetry were undertaken. Those with GORD were randomised. No withdrawals. After end of trial (12 weeks), open study performed using omeprazole (one month) Randomisation by computer but allocation method not described. Compliance monitored by pill counts after treatment period All completed trial. Analysis by intention to treat possible

Participants 17 participants randomised from 71 patients screened. Patients with chronic cough enrolled from newspaper advertisements and outpatients (most were enrolled from community). Mean age and duration of cough in those randomised were not given. Median age of 36 patients in study prior to results of pHmetry was 58 years with cough duration of 5 years (interquartile range 2-10) Inclusion: Aged 18-80 years with chronic cough (>= 6 weeks) with score of >= 2 and with excessive acid reflux distally or proximal (defined by pHmetry) Exclusion: Abnormal chest radiology, positive methacholine test, smokers, ex-smokers =< 3 months or >= 20 pack-yr history of smoking, upper respiratory infection within 8 weeks of entry, use of angiotensin converting enzyme inhibitor or beta blockers, unable to be weaned off some medications (corticosteroids, methylxanthines, cough suppressants, beta agonist, anti-cholinergics, or anti- inflammatory agents), history of pulmonary disease, malignancy, or any co-morbid condition requiring treatment

Interventions Omeprazole 40 mg bd or identical appearing placebo for 12 weeks

Outcomes Cough severity and cough frequency scale (0-8), day and night, measured daily by diary cards for 12 weeks Response to treatment deﬁned as weekly cough frequency combined with severity score for daytime or nighttime cough of =< 1 for >= 2weeks consecutively

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 31 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Ours 1999 (Continued)

Notes Paper provided failure and success rates but no details on cough scores that can be entered for other meta- analysis Open trial results not used for analysis

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Pawar 2007

Methods Single centre, double-blinded, randomised placebo-controlled trial that compared twice-daily rabeprazole with placebo for 90 days in adults with “post nasal drip (PND), throat clearing or excessive throat mucus” Subjects were recruited from otolaryngology clinic at Medical College of Wisconsin or through newspaper advertisements. Rigid nasal endoscopy, two-site 24-hour pharyngeal pH probe monitoring (pH threshold =5) and video laryngoscopy performed before Rx. Video laryngoscopy and other 2 outcomes below repeated on day 90 Non prescription antacids (including H2 antagonists and other PPIs), anti-histamines, decongestants disallowed to be used concomitantly Randomisation method and allocation method not described. Jadad score: 3, High quality score B, A, B, B (no intention to treat analysis) Study was ﬁnancially supported in part by a pharmaceutical company

Participants 180 people screened. Of these 53 subjects enrolled, 47 (89%) completed study Rabeprazol group: n=21 (11 males, 10 female), mean age= 55.6 years (range 40-75). Placebo group: n= 26 (12 males, 14 female), mean age=54.3 years (range 24-76) Inclusion: Aged 18-80 with main complain of PND, throat clearing or excessive throat mucous Exclusion: Acute or chronic sinus disease (using nasal endoscopy and/or coronal sinus CT scan), acute rhino sinusitis, nasal polyps or neoplasms, subjects with vasomotor rhinitis or GER treated in preceding 2 months, smoking, oesophageal or gastric surgery, cardiovascular diease, allergies or pregnancy

Interventions Rabeprazole (20 mg, orally twice daily) or placebo for 90 days

Outcomes Visual analogue scales for PND symptoms, reflux symptom index (higher score = worse), and reflux finding score (RFS) (higher score = worse)

Notes Symptomatic score signiﬁcantly improved in intervention arm but the objective score (RFS) was signiﬁ- cantly worse. Wrote to authors 30th May 2008

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Methods Single centre randomised double blind, parallel comparison of rabeprazole vs placebo in adults with chronic cough and GORD (deﬁned on pHmetry). Potential participants screened for exclusion criteria. Patients completed underwent video strobo-laryngoscopy and undertook questionnaire. 24-hour dual probe pHmetry was initially part of protocol but was later optional (as patients poorly tolerated it) Randomisation by computer generated random number table and allocation concealed (using sealed envelopes). Compliance to PPI/placebo not described Analysis available for 30 of the 42 (71.4%) subjects randomised. Intention to treat data not available

Participants 194 patients screened, 42 randomised. Patients enrolled from an academic otolaryngology practice at the University of Cincinnati, USA and from poster and newspaper advertisements. Rabeprazole group (n = 21): mean age (SD) = 52.8 (11.5), 23% males. Placebo group (n = 21): mean age (SD) = 45.8 (11.2), 33. 3% males Inclusion criteria: 1) age > 18 years; 2) history of hoarseness, throat clearing, non-productive cough, globus sensation, or sore throat >4 weeks; and 3) physical examination consistent with diagnosis of laryngopharyngeal reﬂux (ie, edema, erythema, and/or pachydermia) Exclusion criteria included: 1) previous surgery for gastroesophageal reﬂux disease, or current gastrostomy tube; 2) history of hypersecretory disorder (ie, Zollinger Ellison); 3) current tracheotomy tube; 4) current or recent (within 1 month) use of a proton pump inhibitor; 5) current or recent (within 1 month) use of a histamine-2 receptor blocker (except over-the-counter use on an as-needed basis <3 times per week); 6) allergy to proton pump inhibitors; 7) current systemic steroid therapy; 8) previous laryngeal or hypopharyngeal neoplasm; 9) previous radiation therapy to neck; 10) endotracheal tube intubation within past 2 months; 11) diagnosis of vocal cord paralysis; 12) diagnosis of active granulomatous disease (laryngeal or pulmonary, including sarcoid, Wegener’s, tuberculosis, histoplasmosis, blastomycosis) requiring systemic therapy; and 13) suspicion for laryngeal neoplasm requiring biopsy for diagnosis

Interventions Rabeprazole 20 mg twice-daily or placebo (identical appearance) for 2 months. All also received lifestyle modiﬁcation instructions minimize acid reﬂux (avoidance of fatty meals, caffeine, alcohol, smoking, and oral intake within 2 hours of lying down or bedtime, as well as the addition of extra pillows to raise the head or bedpost blocks to elevate the head of the bed 6 inches)

Outcomes The primary outcome measures: change in total reflux symptom scores and proportion of subjects noting significant global improvement. The reflux symptom scores were modification of a previously validated GERD outcome questionnaire that included laryngo-pharyngeal symptoms. The questionnaire included 9 items (heartburn, acid regurgitation, dysphagia, globus, throat clearing, hoarseness, sinus drainage or phlegm, sore throat, and dry cough). Patients responded to frequency of symptoms (never, monthly, weekly, several times per week, or daily) and severity of symptoms (none, mild, moderate, severe, very severe), both scored 0 to 4 on a Likert scale. The total reflux symptom score included sum of frequency and severity scores for all items The secondary outcome measures were change in SF36, change in laryngeal grading of videostrobolaryn- goscopy, and changes in component reflux symptoms scores (frequency, severity, “typical,” “laryngeal,” “pharyngeal,” and individual symptoms). The component reflux symptom scores were defined as follows: frequency = sum of all reflux symptom frequency scores; severity = sum of all reflux symptom severity scores; “typical” = sum of frequency and severity scores for symptoms of heartburn and acid regurgitation; “laryngeal” = sum of frequency and severity scores for symptoms of hoarseness, dry cough, and throat clearing; and “pharyngeal” = sum of frequency and severity scores for symptoms of throat clearing, globus sensation, and phlegm The tertiary outcome measures were changes in frequency of lifestyle factors related to reflux derived from the lifestyle questionnaire. The lifestyle questionnaire included 9 items (smoking, alcohol use, caffeine use, head of bed elevation with blocks, use of 2 pillows to raise head, oral intake within 2 hours of bedtime,

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 33 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Steward 2004 (Continued)

oral intake within 2 hours of lying down, antacid use, and histamine-2 receptor blocker use). Response choices included: never, monthly, weekly, several times per week, and daily (0-4 Likert scale)

Notes Cough reported separately in paper. Authors contacted for further data

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Yes Sealed envelopes concealed randomisation process from study investigators

Vaezi 2006

Methods Multicenter, randomized, placebo controlled parallel group trial that compared twice-daily esomeprazole 40 mg with placebo for 16 weeks in adults with suspected chronic laryngeal symptoms and signs from GOR (but those with significant GI manifestations of GOR were excluded- see exclusion criteria) Screening laryngoscopy and screening period were performed. Eligible patients then underwent a 7- to 14-day screening period without treatment, during which they completed a daily diary card on which they assessed each of 5 symptoms (throat clearing, cough, globus, sore throat, and hoarseness) over the past 24 hrs on a 7-point Likert scale (0 = none to 6 = very severe). At the end of this period, patients each identified their single most bothersome symptom as the primary symptom. Patients eligible for randomization must have completed at least 80% of diary entries, had a cumulative primary symptom score of >= 9, and had >= 3 days with moderately severe symptoms (>= 3 points) over any 7 consecutive days of the 7 to 14 days Participants randomized in a 2:1 ratio (esomeprazole:placebo) at each center by blocks of six. After randomization and before receipt of study medication, patients had for ambulatory pharyngoesophageal pH monitoring (APEM). Patients who refused APEM were allowed top articipate. APEM consisted of two-catheter, three-probe 24-hr study with probe placement in the hypopharynx, proximal esophagus, and distal esophagus. APEM was assessed by a single, independent investigator at a central laboratory, blinded to patient-identifying data. Patients used daily diary card used to assess symptoms each day throughout the 16-week trial. At weeks 0, 4, 8, 12, and 16, the investigator assessed the same symptoms that the patients assessed in their diary cards using a four-point Likert scale (0 = none to 3 = severe). At the final visit, the investigator also provided an overall evaluation of improvement Allocation method not described. Compliance monitored by pill counting. Analysis was by intention to treat not relevant. Total of 17 (11.7%) withdrawals

Participants 146 participants randomised but one dropped out prior to any Rx (number screened not given). Patients enrolled from 7 US centres. Esomeprazole group (n = 95): mean age (SD) = 51.5 (15.2), 48 males and 47 females. Placebo group (n = 50): mean age (SD) = 50.5 (14.5), 23 males and 27 females Inclusion criteria: aged > 18 y and had one or more of the following symptoms for 3 or more consecutive months before screening: throat clearing, cough, globus, sore throat, or hoarseness; and laryngoscopic signs consistent with reflux (assessed by a panel via consensus) Exclusion: clinically significant conditions that might, in the judgment of the investigator, put the patient at risk, influence the results, affect the patient’s ability to participate in the study, or necessitate surgery

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 34 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Vaezi 2006 (Continued)

during the study. Patients who had experienced moderate-to-severe heartburn 3 or more days per week over a 3-month period before screening; history of respiratory or gastro-intestinal malignancy; radiation therapy to the head and neck, lung, or gastrointestinal tract; gastroesophageal surgery; chronic sinusitis or rhinitis in the last year; an allergic cause of laryngitis; an acute traumatic event near the larynx in the last year; tracheostomy or other signiﬁcant laryngeal or tracheal surgery; and substance or alcohol abuse in the past year; other malignancy (except superﬁcial basal cell carcinoma) within the previous 5 years; presence of an infectious cause of laryngitis in the past 3 months; need for continuous therapy within 1 week of randomisation with diazepam, phenytoin, mephenytoin, warfarin, anticholinergics, antineoplastics, prostaglandin analogs, H2-receptor antagonists, steroids (inhaled, oral, or intravenous), promotility drugs, and sucralfate; use of tobacco products in the past year, any PPI in the last 2 weeks, theophylline, or any other investigational compound or participation in another investigational drug study in the past 30 days; and any contraindication to esomeprazole, such as known or suspected allergy or sensitivity to any PPI; or pregnancy and lactating women

Interventions Esomeprazole 40 mg twice-daily or placebo (identical appearance) for 16 weeks

Outcomes Patients used daily diary card used to assess symptoms each day throughout the 16-week trial. At weeks 0, 4, 8, 12, and 16, the investigator assessed the same symptoms that the patients assessed in their diary cards using a 4-point Likert scale (0 = none to 3 = severe). At the final visit, the investigator also provided an overall evaluation of improvement for each symptom as follows: 0 = resolved, 1 = improved, 2 = same, 3 = worse Laryngoscopic scoring performed at week 8 and 16. The primary outcome was % patients who had resolution of the primary symptom according to patient diary card assessment at the final visit. Symptom resolution was defined as a primary symptom severity score of 0 (none) during the last 7 days of the study, but allowing a score of 1 (minimal severity) for up to 3 days Secondary efficacy variables included relief of the primary symptom (by patient ), % symptom-free days, and % of symptom-free patients, symptom severity change from baseline, and the investigator’s assessment of symptom-free patients. Relief of the primary symptom was defined as a greater than 50% reduction from baseline in the primary symptom score during the final week

Notes Authors provided additional cough speciﬁc data.

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Methods Multi-centre randomised double blind, parallel study comparing a pre-thickened formula with placebo in infants with regurgitant reﬂux Parents of infants kept diary cards for 2 days and if eligible randomised. After one week, infants were reviewed and commenced on pharmacological intervention (ranitidine after day 7/8 and cisapride after day 14/15). Further follow up till day 35/36 Randomisation well described based on site but allocation method not described Compliance monitoring done. Dropouts (n = 12, 10.9% of those randomised. Dropouts included in some analysis but not described as treatment failures. Analysis by intention to treat not possible

Participants 110 infants randomised (number screened not given) from 6 North American paediatric centres. 7 drop outs and of remaining 103, mean age of 55 (27males, 28 females) intervention group was 61 days (SD 4) and that of control group (26males, 23 females) was 58 days (SD 4) Inclusion: >= 5 regurgitations per day for 2 baseline days, age 14-120 days, gestational age > 37 weeks, birth weight >= 2.5kg, maternal age >= 18 years Exclusion: congenital abnormalities interfering with normal feeding or causing repeated regurgitation. Fever, infectious illness, clinical diagnosis of milk or soy protein allergy, complicated GORD (oesophagitis, haematemesis, recurrent respiratory symptoms, failure to thrive), previous treatment with thickened formula, or treatment with prokinetic medications 5 days before start of study

Interventions A pre-thickened formula (Enfamil AR) with placebo for 35/36 days, with equal access to additional pharmacological intervention

Outcomes 1. Regurgitation frequency, regurgitation volume, as documented on daily diary sheets for the ﬁrst week and then 2 days/week subsequently. 2. Percentage feeds followed by choke/gag/cough. 3. Trouble sleeping 4. Adverse events Measures at week 1 and end of participation.

Notes Authors replied but were unable to provide information on cough alone as an outcome measure

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Information not available

Methods Single centre, double-blinded, randomised placebo-controlled trial that compared once daily pantoprazole with placebo for 12 weeks in adults Protocol was 2-wk run-in, 12-wk randomised treatment, and 4-wk off-treatment follow-up periods. During the run-in period, subjects ﬁlled out weekly diaries. Subjects were seen mid protocol. Naso- pharyngoscopy, oesophageal manometry and triple sensor pH monitoring performed performed before Rx. Naso-pharyngoscopy, laryngoscopy and triple sensor pH monitoring performed repeated on end of week 12. Diaries collected till 4 weeks post Rx Randomisation method and allocation method clearly described Jadad score: 5, High quality score: A, A, A, A Study was ﬁnancially supported by a pharmaceutical research grant

Participants 42 of 51 who had pH monitoring qualified for study. Of these 42 adults, 39 were randomised and all completed study Pantoprazole group: n=20 (7 males, 13 female), mean age= 39 years (range 23-59). Placebo group: n=19 (6 males, 13 female), mean age=37 years (range 23-61) Subjects were recruited from otolaryngology clinic at University of Louisville or through newspaper advertisements Inclusion criteria: Subjects with the main complaint of hoarseness, sore throat, throat burning/pain, throat clearing, voice loss, cough, excessive throat mucus, globus, or choking for >3 days per wk in the past 2 months. The diagnosis of LPR was confirmed by laryngeal exam and a positive pH test of the hypopharynx or distal oesophagus (defined as total number of hypopharyngeal reflux episodes >=3 or distal oesophageal total time ph< 4 was >=4.2%) Exclusion criteria: Previous treatment of LPR or GORD, known gastroparesis, connective tissue disorder, previous endoscopic or surgical antireflux procedure, or gastric surgery. Also those with vocal cord ulcer, vocal cord granuloma, or laryngeal malignancy seen on naso-pharyngoscopic examination were excluded

Interventions Pantoprazole (40 mg once daily) or placebo for 12 weeks

Outcomes Symptom assessment: weekly diaries of visual analogue scales (0: none to 20: severe) for six laryngeal symptoms: globus, cough, sore throat, hoarseness, throat clearing, and excessive throat mucus. A total laryngeal symptom score (0-120) was defined as the sum of the laryngeal symptoms. A similar visual analogue scale (0-20) was obtained for daytime and nighttime heartburn Nasopharyngoscopy: Reflux finding score (RFS) (higher=worse)

Notes

Risk of bias

Item Authors’ judgement Description

Allocation concealment? Unclear Randomisation assignment generated by third party in blocks of 4 with sequential numbering but insuf- ﬁcient details to be certain if allocation concealment occurred

APEM: ambulatory pharyngoesophageal pH monitoring; bd: twice daily; GOR: Gastroesophageal reﬂux; GORD: Gastroesophageal reﬂux disease; SD: Standard deviation; vs: versus

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 37 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Characteristics of excluded studies [ordered by study ID]

Study Reason for exclusion

Ahmad 2004 Non RCT.Prospective follow up study of 303 adults with otolaryngeal manifestations of GORD. 86% had cough and choking events presumed secondary to GORD (diagnosed on gastroscopy) were treated with antireﬂux therapy (PPI’s in 90%, ranitidine in 3%, Gaviscon in 11%). Complete response to therapy occurred in 25.7%, no response in 23.1% and the rest had variable response. Adverse events not mentioned

Allen 1998 Non RCT. Prospective follow up study of 195 adults with GORD (diagnosed objectively by combination of gastroscopy, pHmetry, manometry) treated with Nissen fundoplication. Cough was primary reason for surgery in 42 (21.5%) patients. Cough scores signiﬁcant reduced post surgery - 51% were cough free and 31% improved. Cough scores correlated to dysphagia score (r = 0.233, P = 0.004) but did not relate to scores for heartburn, pain or regurgitation. Adverse events not mentioned

Allen 2002 Non RCT. Prospective study on predictors for response of cough to laparoscopic fundoplication in 354 adults (sub group of total 603 fundoplication). Cough was eliminated in 54% of the 81% (287) followed-up and improved in further 31%. Strongest predictor for response of cough was preoperative cough score (r = 0.62). Change in cough score on and off PPI ie positive response to PPI (r = 0.296) was a weak predictor. Time scale of response of PPI was not provided. Those with positive Bernstein test were more likely (P = 0.024) to respond but predictive value not given. Adverse events not reported

Allen 2004 Non RCT. Follow up of 209 with respiratory symptoms, mainly cough who underwent laparoscopic Nissen fundoplication. 81% were followed at 6 months, 73% at 2 years, and 60% at 5 years. Cough improved in 83% at 6 months, 74% at 2 years, and 71% at 5 years

Baldi 2006 Placebo not used. RCT comparing 30 mg bd of lansoprazole to 30 mg daily for 12 weeks. Authors suggest a positive response at 4 weeks of PPI is an effective criteria for assessing response to PPI treatment

Baldi 2006a Non RCT. Observational prospective study.

Belafsky 2008 Retrospective study on LPR

Brouwer 2003 Non RCT. Prospective study of 29 adults with GORD (diagnosed on gastroscopy) and respiratory symptoms including cough (n = 19) following laparoscopic Nissen fundoplication. 14 months follow up; 3 required conversion to open procedure, 4 with major complications. Cough (assessed by subjective score) reported to resolve completely in 81% and improved in 13%. pHmetry ﬁndings in those with respiratory symptoms similar to those with gastrointestinal symptoms

Chandra 2007 Review paper. No additional references identiﬁed.

Chen 2000 Non RCT. Retrospective and prospective study of 90 adults undergoing laparoscopic Nissen fundoplication, 97% with typical GI symptoms, 56% with concurrent non-GI symptoms including cough in 17%. Post surgery patients with non-GI symptoms improved in 54% but reﬂux symptoms in 95%. New symptoms of ﬂatulence, belching, dysphagia, chest and abdominal pain were reported but not measured. Surgical complications not mentioned

Coron 2007 Review paper. One additional paper identiﬁed from references

Dalby-Payne 2003 Meta-analysis of cisapride for treatment of GORD in children. Cough was not an outcome measure in meta- analysis

DeMeester 1990 Non RCT. Prospective study evaluating the effect of fundoplication in 17 adults with chronic/persistent respiratory symptoms (including cough) and GORD in-patients with or without chest radiograph abnormality. Surgery was not beneficial only on those whose respiratory symptoms occurred pre reflux event. Those whose respiratory symptoms occurred during or within 3 mins following or were unrelated to reflux episodes and had pre surgical normal motility were more likely to benefit

Dore 2007 None placebo controlled RCT. Study used different PPIs (rabeprazole, pantoprazole, esomeprazole and lansoprazole) and placebo was not used

Duffy 2003 Non RCT. Prospective study of 148 adults with GORD (diagnosed on gastroscopy, pHmetry or manometry) undergoing laparoscopic Nissen fundoplication. Of these 31% had cough pre surgery. Modiﬁed Gastrointestinal Quality of Life Index used and no cough speciﬁc scores utilised. Improved QOL post surgery reported, surgical complications not reported

Ekstrom 2000 Non RCT. Prospective study evaluating the effect of fundoplication on 24 patients with severe GORD) and concomitant asthma (n = 13) or chronic cough (n = 11). In non-asthmatic patients, daytime cough reduced by 47% and night cough by 80% 12 months after surgery

El Hennawi 2004 Non RCT. 64 of 80 adults treated with ’anti-reﬂux Rx’. Before and after Rx effect described

Eubanks 2001 Non RCT. Before and after PPI study for 3 months relating symptoms of cough, hoarseness and throat clearing with proximal pHmetry in 14 adults, 7 of whom had cough. 5 of the 7 who improved with PPI had reduction of > 70% of pharyngeal acid reﬂux ie 2 of 7 improved without signiﬁcant reduction. PPI well tolerated, no side effects reported

Farrell 2001 Prospective non RCT. Symptoms measured pre and post fundoplication (at 6 weeks) in 324 adults of which 67 had cough. Post surgery, cough resolved in 67%, those with atypical symptoms of GORD were more likely to have had less severe pHmetry indices than those with classical GORD symptoms

Fock 2008 Consensus guidelines. No additional references identiﬁed.

Fraser 2000 Prospective non RCT. Symptoms measured pre and post PPI treatment (20 mg daily and if no response by 3 months increased to 40 mg daily) in 87 adults (25 had cough as primary symptom). After 6 weeks, 52% had good response (undeﬁned)

Gatta 2007 Meta-analysis assessing the effectiveness of medical or surgical therapy for reﬂux disease in adult patients with laryngeal or pharyngeal symptoms presumed to be due to gastro-oesophageal reﬂux disease The authors concluded that “Therapy with a high-dose proton pump inhibitor is no more effective than placebo in producing symptomatic improvement or resolution of laryngo-pharyngeal symptoms”

Greason 2002 Retrospective non RCT in adults with primary respiratory symptoms. Most had abnormal pulmonary function test

Grill 1985 Non RCT. Effect of domperidone (0.3-0.6 mg/kg/dose qid) on symptoms and oesophageal and gastric motility in 15 infants (six boys, mean age 7.9 months), with moderate to severe GORD and upper gastrointestinal motility disturbances. Vomiting, “spitting,” and coughing improved signiﬁcantly after 6 weeks therapy

Habermann 1999 Non RCT. 6-week PPI (pantoprozole) used in 29 adults with ’laryngeal’ disorders presumed secondary to GORD with symptoms including cough. Follow-up at 6 weeks at end of treatment and at 3 months while off therapy. Signiﬁcant change in symptoms of heart-burn, hoarseness, sore throat and dyspnoea but no signiﬁcant reduction in cough reporting at 6 weeks

Habermann 2002 Non RCT. Prospective study of 14 adults patients with pharyngoesophageal gastric acid reflux (symptoms include cough) receiving pantoprazole, 40 mg daily for 6 weeks. Significant improvement was observed in all patients at 6 weeks, in 71% of the patients the 24-hour pH-metry indicated reflux

Hui 2000 Retrospective non RCT- 30 adults with cough and GORD. 60% had symptoms resolve post fundoplication assessed at 6-36 months. Those without pre-operative nausea were more likely to have resolution of cough

Hunter 1996 Non RCT. Review of 300 adults post laparoscopic fundoplication (252 Nissen and 48 Toupet) for gastroesophageal reflux refractory to medical therapy or requiring daily omeprazole or high-dose H2 antagonists. 64% had atypical gastroesophageal reflux symptoms (including cough, asthma, hoarseness, chest pain) pre surgery. Minor complications occurred in 6% and major complications in 2%. Atypical reflux symptoms were eliminated or improved in 87% and no better in 13%

Irwin 1993 Non RCT. Determination of characteristics of 12 adults with cough and GORD (diagnosed on disappearance or marked improvement or treated with high protein, low fat, anti reﬂux diet). 75% of cohort had otherwise silent GORD and pHmetry sensitive in 92% but positive Bernstein and Baker tests unhelpful

Irwin 2002 Non RCT. Report on 8 adults said to have medication resistant GORD (objective diagnosis) and cough, 4 of whom had previous fundoplication. All had surgery or repeat fundoplication and cough scores reduced post operative (median of 41 days). One year post fundoplication/repeat fundoplication, although median cough scores of group were still signiﬁcantly lower than immediate post surgery, cough scores increased in 5 patients. Surgical complications not mentioned

Issing 2004 Non RCT. 22 adults with chronic cough, globus pharyngeus, hoarseness or dysphonia (from ENT clinics) treated with esomeprazole. After 8 weeks, 95% were symptom free

Katzka 1996 Non RCT. Prospective study of 14 adults with persistent symptoms (typical GI and atypical-asthma, cough, hoarseness, throat burning or undetermined chest pain) despite standard PI therapy (omeprazole 20 mg bd). Of the 6 with atypical symptoms studied, increased dose of omeprazole (20 mg qid) controlled reflux but did not control atypical symptoms and pHmetry showed poor correlation between symptoms and acid reflux (defined as > 50% association of symptoms with pH < 4 within 5 mins)

Leeder 2002 Non RCT. Prospective study on effect of laparoscopic fundoplication in 14 adults objective evidence of GORD (without oesophagitis on endoscopy or pHmetry). Patients grouped into typical GI (heartburn, volume reﬂux, vomiting or regurgitation) or atypical (cough, chest pain, bloating, belching, or choking). 3 of the 7 (43%) in atypical symptom group had poor outcome post surgery. 2 of these patients had cough but the effect of these 2 patients were not explicitly reported

Monini 2006 Non placebo controlled trial. All with laryngeal symptoms clinically graded as having GORD were given PPI

Murray 2006 Non RCT. Study described the outcome of patients with cough and paradoxical vocal fold movement disorder treated with respiratory retraining therapy and management of laryngopharyngeal reﬂux

Novitsky 2002 Non RCT. Prospective study on effect of fundoplication in 21 adults with chronic cough and GORD and were unresponsive to PPIs and other medical therapy. Fundoplication was laparoscopic in 18 and open in 3. Cough improved in 86% of which 62% had complete resolution of cough. No major perioperative complications occurred

Omari 2006 RCT on effect of baclofen on physiological parameters of GER with no clinical correlates measured as an outcome, as reported by correspondence to corresponding author

Oridate 2008 Non RCT. Observational study on forty-ﬁve patients with laryngopharyngeal and esophageal symptoms receiving continuous proton-pump inhibitor for up to 6 months

Poe 2003 Non RCT. Prospective study on 56 adults with chronic cough presumed related to GORD (clinical diagnosis) . Patients treated with PPI and if no response, metoclopromide or cisapride added. In 86%, cough eliminated or markedly improved after 4 weeks

Songur 2008 Non-RCT examining effect of PPI in adults with symptomatic cough and GERD diagnosed using scintigraphy

Swoger 2006 Prospective controlled non-randomised trial of 25 adults with LPR (of which 30-31% had chronic cough) to fundoplication vs no fundoplication. No difference found between groups at one year, despite technical success (increased lower oesophageal sphincter pressure and normalised oesophageal acid) in the fundoplication group

Thoman 2002 Non RCT. Retrospective study on effect of laparoscopic fundoplication 37 adults with cough and GORD. Complete resolution of cough occurred in 64% with 10% reported improvement. Patients with cough and GORD were more likely to have dysphagia post surgery

Tibbling 1993 Non RCT. Comparison of adults (mixed respiratory and non-respiratory disease) with GORD treated with omeprazole (n = 89) and fundoplication (n = 199). Omeprazole (mean 50 days) did not signiﬁcantly relieved cough (34% pre and 28% post), but patients who had fundoplication and crural repair showed a highly signiﬁcant reduction of cough (34% pre and 7% post)

Tibbling 1995 Similar data set to 1993 paper. van Zanten 2006 RCT on 224 adults with functional dyspepsia using esomeprazole vs placebo. Cough was not listed amongst the symptoms examined

Waring 1995 Non RCT. Prospective study evaluating the effect of anti-reﬂux therapy (omeprazole 20-40 mg/day in 6 and fundoplication in 21) in 27 adults with GORD and chronic cough or hoarseness. 80% improved including 36% who became asymptomatic. Improvement of cough was more likely in those whose heartburn was eliminated. No features of dual pHmetry could predict responders vs non- responders

Wo 1997 Non RCT. Prospective study of 22 adults with posterior laryngitis (4 of whom the major symptom was cough) treated with 40 mg qid of omeprazole for 8 weeks. Symptom score for cough signiﬁcantly reduced by 8 weeks

Wright 2003 Non RCT. Prospective study on effect of laparoscopic fundoplication in 145 adults with hiatus hernia and symptoms suggestive of GORD and laryngopharyngeal reﬂux. 55% had abnormal pHmetry. 41% of cohort had cough pre-surgery of which 92% reported improvement. 5 (1.4%) had signiﬁcant complications (2nd surgery for bleeding, oesophageal perforation, severe odynophagia required in 3 and pulmonary embolus in 2)

Xinias 2003 Open placebo controlled but non-randomised trial using modified milk for GORD in 56 infants. After 4 weeks of therapy, there was no change in cough scores although significant reduction in reflux index, vomiting and regurgitation episodes were found in treatment arm compared to placebo arm

Yang J 2006 Non RCT. Subjects with chronic treated with 8 weeks of PPI.

ENT: Ear nose throat GI: GORD: Gastroesophageal reﬂux disease PI: PPI: qid: QOL: Quality of life RCT: Randomised controlled trial Rx: Treatment

Characteristics of studies awaiting assessment [ordered by study ID]

Morice 2008

Methods RCT on proscriptive or standard anti-reﬂux advice with brief dietary weight loss intervention

Participants Obese adults with presumed cough associated with reﬂux

Interventions Dietary

Outcomes Cough QOL

Notes

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 42 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. DATA AND ANALYSES

Comparison 1. Thickened versus unthickened feeds (infants)

No. of No. of Outcome or subgroup title studies participants Statistical method Effect size

1 Subjects cured (of cough) at end 1 Odds Ratio (M-H, Fixed, 95% CI) Totals not selected of study 2 Cough frequency 1 Coughs/hr (Fixed, 95% CI) Totals not selected

Comparison 2. PPI versus placebo (> 18 years)

No. of No. of Outcome or subgroup title studies participants Statistical method Effect size

1 Clinical failures (still coughing at 4 191 Odds Ratio (M-H, Random, 95% CI) 0.46 [0.19, 1.15] end of trial or reporting period) 1.1 Medical clinics based 2 38 Odds Ratio (M-H, Random, 95% CI) 0.17 [0.02, 1.73] enrolment 1.2 Otolaryngology based 2 153 Odds Ratio (M-H, Random, 95% CI) 0.56 [0.21, 1.49] enrolment 2 Mean cough score at end of 4 109 Std. Mean Difference (IV, Fixed, 95% CI) -0.38 [-0.77, 0.00] trial (1st arm crossover/parallel group trials) 2.1 Medical clinic based 1 21 Std. Mean Difference (IV, Fixed, 95% CI) -0.86 [-1.77, 0.05] enrolments 2.2 Otolargyngology based 3 88 Std. Mean Difference (IV, Fixed, 95% CI) -0.28 [-0.70, 0.14] enrolments 3 Change in cough scores 6 161 Std. Mean Difference (IV, Fixed, 95% CI) -0.39 [-0.71, -0.08] (end-beginning of intervention - 1st arm crossover/parallel group trials) 3.1 studies using omeprazole 2 51 Std. Mean Difference (IV, Fixed, 95% CI) -0.71 [-1.29, -0.14] 3.2 studies using other PPI 4 110 Std. Mean Difference (IV, Fixed, 95% CI) -0.26 [-0.64, 0.12] (lansoprazole, pantoprazole, rabeprazole) 4 Change in cough scores 2 SD units (Fixed, 95% CI) -0.41 [-0.75, -0.07] (crossover studies; standardised scale) 4.1 Sub-category 2 SD units (Fixed, 95% CI) -0.41 [-0.75, -0.07] 5 Absolute cough scores (crossover 2 SD units (Fixed, 95% CI) -0.29 [-0.62, 0.04] studies, standardised scale) 6 Change in cough score after 4 2 51 Std. Mean Difference (IV, Fixed, 95% CI) -0.51 [-1.08, 0.06] weeks treatment (1st arm cross over/parallel group trials)

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 43 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. 7 Difference in cough scores at 2 51 Std. Mean Difference (IV, Fixed, 95% CI) -0.44 [-1.04, 0.16] week 8 - week 4 (1st arm cross over/parallel group trials)

Analysis 1.1. Comparison 1 Thickened versus unthickened feeds (infants), Outcome 1 Subjects cured (of cough) at end of study.

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 1 Thickened versus unthickened feeds (infants)

Outcome: 1 Subjects cured (of cough) at end of study

Studyorsubgroup Thickenedfeeds Unthickenedfeeds OddsRatio OddsRatio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Chao 2007 0/5 2/4 0.09 [ 0.00, 2.68 ]

0.01 0.1 1 10 100 Favours thickened feeds Favours control

Analysis 1.2. Comparison 1 Thickened versus unthickened feeds (infants), Outcome 2 Cough frequency.

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 1 Thickened versus unthickened feeds (infants)

Outcome: 2 Cough frequency

Studyorsubgroup Coughs/hr(SE) Coughs/hr Coughs/hr IV,Fixed,95% CI IV,Fixed,95% CI Orenstein 1992 1.1 (0.89) 1.10 [ -0.64, 2.84 ]

-4 -2 0 2 4 Thickened better Unthickened better

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 44 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Analysis 2.1. Comparison 2 PPI versus placebo (> 18 years), Outcome 1 Clinical failures (still coughing at end of trial or reporting period).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 1 Clinical failures (still coughing at end of trial or reporting period)

Studyorsubgroup PPI Placebo OddsRatio Weight OddsRatio M- M- H,Random,95% H,Random,95% n/N n/N CI CI

1 Medical clinics based enrolment Kiljander 2000 7/9 12/12 8.2 % 0.12 [ 0.01, 2.85 ]

Ours 1999 7/8 9/9 7.4 % 0.26 [ 0.01, 7.43 ] Subtotal (95% CI) 17 21 15.6 % 0.17 [ 0.02, 1.73 ] Total events: 14 (PPI), 21 (Placebo) Heterogeneity: Tau2 = 0.0; Chi2 = 0.11, df = 1 (P = 0.74); I2 =0.0% Test for overall effect: Z = 1.49 (P = 0.14) 2 Otolaryngology based enrolment Eherer 2003 2/5 4/6 13.5 % 0.33 [ 0.03, 3.93 ]

Vaezi 2006 79/94 43/48 70.9 % 0.61 [ 0.21, 1.80 ] Subtotal (95% CI) 99 54 84.4 % 0.56 [ 0.21, 1.49 ] Total events: 81 (PPI), 47 (Placebo) Heterogeneity: Tau2 = 0.0; Chi2 = 0.20, df = 1 (P = 0.66); I2 =0.0% Test for overall effect: Z = 1.17 (P = 0.24) Total (95% CI) 116 75 100.0 % 0.46 [ 0.19, 1.15 ] Total events: 95 (PPI), 68 (Placebo) Heterogeneity: Tau2 = 0.0; Chi2 = 1.14, df = 3 (P = 0.77); I2 =0.0% Test for overall effect: Z = 1.66 (P = 0.097)

0.005 0.1 1 10 200 PPI better Placebo better

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 45 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Analysis 2.2. Comparison 2 PPI versus placebo (> 18 years), Outcome 2 Mean cough score at end of trial (1st arm crossover/parallel group trials).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 2 Mean cough score at end of trial (1st arm crossover/parallel group trials)

Std. Std. Mean Mean Studyorsubgroup PPI Placebo Difference Weight Difference N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI

1 Medical clinic based enrolments Kiljander 2000 9 5.89 (5.01) 12 9.91 (4.1) 17.8 % -0.86 [ -1.77, 0.05 ] Subtotal (95% CI) 9 12 17.8 % -0.86 [ -1.77, 0.05 ] Heterogeneity: not applicable Test for overall effect: Z = 1.84 (P = 0.065) 2 Otolargyngology based enrolments Eherer 2003 5 0.4 (0.55) 6 1.5 (1.64) 9.3 % -0.79 [ -2.04, 0.47 ]

Noordzij 2001 15 243.3 (346.62) 15 328.2 (473.67) 28.6 % -0.20 [ -0.92, 0.52 ]

Pawar 2007 21 0.29 (0.64) 26 0.5 (1.1) 44.3 % -0.22 [ -0.80, 0.35 ] Subtotal (95% CI) 41 47 82.2 % -0.28 [ -0.70, 0.14 ] Heterogeneity: Chi2 = 0.71, df = 2 (P = 0.70); I2 =0.0% Test for overall effect: Z = 1.29 (P = 0.20) Total (95% CI) 50 59 100.0 % -0.38 [ -0.77, 0.00 ] Heterogeneity: Chi2 = 1.98, df = 3 (P = 0.58); I2 =0.0% Test for overall effect: Z = 1.95 (P = 0.051) Test for subgroup differences: Chi2 = 1.27, df = 1 (P = 0.26), I2 =21%

-4 -2 0 2 4 Favours PPI Favours Placebo

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 46 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Analysis 2.3. Comparison 2 PPI versus placebo (> 18 years), Outcome 3 Change in cough scores (end- beginning of intervention - 1st arm crossover/parallel group trials).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 3 Change in cough scores (end-beginning of intervention - 1st arm crossover/parallel group trials)

Std. Std. Mean Mean Studyorsubgroup PPI Placebo Difference Weight Difference N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI

1 studies using omeprazole Kiljander 2000 9 -4.55 (5.18) 12 1.33 (5.18) 11.4 % -1.09 [ -2.03, -0.15 ]

Noordzij 2001 15 -112.9 (165.68) 15 -20 (207.16) 19.0 % -0.48 [ -1.21, 0.25 ] Subtotal (95% CI) 24 27 30.4 % -0.71 [ -1.29, -0.14 ] Heterogeneity: Chi2 = 1.01, df = 1 (P = 0.32); I2 =1% Test for overall effect: Z = 2.42 (P = 0.015) 2 studies using other PPI (lansoprazole, pantoprazole, rabeprazole) Eherer 2003 5 -1.4 (2.7) 6 -2.66 (3.14) 6.9 % 0.39 [ -0.81, 1.59 ]

Havas 1999 8 -0.9 (3.37) 7 -0.86 (3.33) 9.8 % -0.01 [ -1.03, 1.00 ]

Pawar 2007 21 -0.52 (1.123) 26 0.08 (0.775) 28.9 % -0.62 [ -1.21, -0.03 ]

Steward 2004 18 -1.2 (2.48) 19 -0.9 (3.02) 24.1 % -0.11 [ -0.75, 0.54 ] Subtotal (95% CI) 52 58 69.6 % -0.26 [ -0.64, 0.12 ] Heterogeneity: Chi2 = 3.01, df = 3 (P = 0.39); I2 =0% Test for overall effect: Z = 1.33 (P = 0.19) Total (95% CI) 76 85 100.0 % -0.39 [ -0.71, -0.08 ] Heterogeneity: Chi2 = 5.68, df = 5 (P = 0.34); I2 =12% Test for overall effect: Z = 2.44 (P = 0.015) Test for subgroup differences: Chi2 = 1.66, df = 1 (P = 0.20), I2 =40%

-2 -1 0 1 2 Favours PPI Favours Placebo

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 47 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Analysis 2.4. Comparison 2 PPI versus placebo (> 18 years), Outcome 4 Change in cough scores (crossover studies; standardised scale).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 4 Change in cough scores (crossover studies; standardised scale)

Studyorsubgroup SDunits(SE) SDunits Weight SDunits IV,Fixed,95% CI IV,Fixed,95% CI

1 Sub-category Eherer 2003 -0.34 (0.277) 38.7 % -0.34 [ -0.88, 0.20 ]

Kiljander 2000 -0.45 (0.22) 61.3 % -0.45 [ -0.88, -0.02 ] Total (95% CI) 100.0 % -0.41 [ -0.75, -0.07 ] Heterogeneity: Chi2 = 0.10, df = 1 (P = 0.76); I2 =0.0% Test for overall effect: Z = 2.37 (P = 0.018) Test for subgroup differences: Not applicable

-1 -0.5 0 0.5 1 Favours PPI Favours placebo

Analysis 2.5. Comparison 2 PPI versus placebo (> 18 years), Outcome 5 Absolute cough scores (crossover studies, standardised scale).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 5 Absolute cough scores (crossover studies, standardised scale)

Studyorsubgroup SDunits(SE) SDunits Weight SDunits IV,Fixed,95% CI IV,Fixed,95% CI Eherer 2003 -0.161 (0.27) 39.9 % -0.16 [ -0.69, 0.37 ]

Kiljander 2000 -0.375 (0.22) 60.1 % -0.38 [ -0.81, 0.06 ] Total (95% CI) 100.0 % -0.29 [ -0.62, 0.04 ] Heterogeneity: Chi2 = 0.38, df = 1 (P = 0.54); I2 =0.0% Test for overall effect: Z = 1.70 (P = 0.089) Test for subgroup differences: Not applicable

-1 -0.5 0 0.5 1 Favours PPI Favours placebo

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 48 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Analysis 2.6. Comparison 2 PPI versus placebo (> 18 years), Outcome 6 Change in cough score after 4 weeks treatment (1st arm cross over/parallel group trials).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 6 Change in cough score after 4 weeks treatment (1st arm cross over/parallel group trials)

Std. Std. Mean Mean Studyorsubgroup PPI Placebo Difference Weight Difference N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI

Kiljander 2000 9 -1.78 (4.47) 12 -1.58 (3.78) 43.1 % -0.05 [ -0.91, 0.82 ]

Noordzij 2001 15 -166.8 (197.17) 15 9.4 (199.98) 56.9 % -0.86 [ -1.62, -0.11 ] Total (95% CI) 24 27 100.0 % -0.51 [ -1.08, 0.06 ] Heterogeneity: Chi2 = 1.95, df = 1 (P = 0.16); I2 =49% Test for overall effect: Z = 1.76 (P = 0.078) Test for subgroup differences: Not applicable

-4 -2 0 2 4 Favours PPI Favours Placebo

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 49 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Analysis 2.7. Comparison 2 PPI versus placebo (> 18 years), Outcome 7 Difference in cough scores at week 8 - week 4 (1st arm cross over/parallel group trials).

Review: Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults

Comparison: 2 PPI versus placebo (> 18 years)

Outcome: 7 Difference in cough scores at week 8 - week 4 (1st arm cross over/parallel group trials)

Std. Std. Mean Mean Studyorsubgroup PPI Placebo Difference Weight Difference N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI

Kiljander 2000 9 -4.67 (2.17) 12 2.16 (3.9) 30.1 % -2.00 [ -3.09, -0.90 ]

Noordzij 2001 15 53.9 (161.51) 15 10.6 (206.83) 69.9 % 0.23 [ -0.49, 0.95 ] Total (95% CI) 24 27 100.0 % -0.44 [ -1.04, 0.16 ] Heterogeneity: Chi2 = 11.09, df = 1 (P = 0.00087); I2 =91% Test for overall effect: Z = 1.45 (P = 0.15) Test for subgroup differences: Not applicable

-4 -2 0 2 4 Favours PPI Favours Placebo

APPENDICES Appendix 1. Archive of search results (All years to April 2007) The Airways Group specialised register/search identified 763 potentially relevant titles in the original search in 2004. After assessing the abstracts, 84 papers were obtained for considered for inclusion into review. The main reason for non-eligibility of studies for review criteria was the non-controlled non-randomised nature of the respective studies (see table ’Characteristics of excluded studies’). One controlled but non-randomised study was also excluded (Xinias 2003). Eleven studies were included in the original version of the review (see ’Characteristics of included studies’), and all but one (Vanderhoof 2003) were single centre studies. The multi-centre study (Vanderhoof 2003) also the only study supported by commercial interest then. In the 2005 update a further 11 articles were retrieved; one study was included (El Serag 2001) and the rest were excluded, with two described in the excluded table (El Hennawi 2004; Issing 2004). Following the 2006 search update, eight papers were retrieved. One RCT in adults (Vaezi 2006) was included and another RCT was excluded (Baldi 2006) as a placebo was not used. The excluded study compared two doses of lansoprazole for adults with chronic cough (Baldi 2006). Following the 2007 search update, 407 abstracts were reviewed, 12 papers were retrieved; none fulfilled eligibility criteria. One study that was excluded but should be highlighted was that of a controlled (but non randomised) trial on fundoplication for laryngopharyngeal reflux (LPR) of which 30-31% of the subjects had chronic cough, the 2nd or 3rd most common symptom (Swoger 2006). Swoger 2006 described no difference between groups at one year, despite technical success of the fundoplication. A systematic review on the efficacy of PPIs for LPR was also excluded (Gatta 2007). The paper (Gatta 2007) however revealed a relevant study that was not found in previous searches; this study (Steward 2004) is now included in this update.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 50 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. Appendix 2. CENTRAL search strategy #1 MeSH descriptor Gastroesophageal Reflux explode all trees in MeSH products #2 ((gastro-oesophageal next reflux) or (gastroesophageal next reflux) or (gastro-esophageal next reflux) or reflux or ger or gerd or acid or esophagus or oesophagus) #3 MeSH descriptor Esophagus explode all trees in MeSH products #4 (#1 OR #2 OR #3) #5 MeSH descriptor Cough explode all trees in MeSH products #6 MeSH descriptor Bronchitis explode all trees in MeSH products #7 cough* #8 bronchiti* #9 (#5 OR #6 OR #7 OR #8) #10 (#4 AND #9)

Appendix 3. MEDLINE search strategy 1. exp COUGH/ 2. exp Bronchitis/ 3. (cough$ or bronchit$).mp. 4. 1 or 2 or 3 5. exp GASTROESOPHAGEAL REFLUX/ 6. exp Esophagus/ 7. (gastro-oesophageal reflux or gastroesophageal reflux or gastro-esophageal reflux or reflux or ger or gerd or acid or esophagus or oesophagus).mp. 8. 5 or 6 or 7 9. 4 and 8

Appendix 4. EMBASE search strategy 1. exp Coughing/ 2. exp Bronchitis/ 3. (cough$ or bronchit$).mp. 4. 1 or 2 or 3 5. exp Gastroesophageal Reflux/ 6. exp esophagus/ 7. (gastro-oesophageal reflux or gastroesophageal reflux or gastro-esophageal reflux or reflux or ger or gerd or acid or esophagus or oesophagus).mp. 8. 5 or 6 or 7 9. 4 and 8

WHAT’S NEW Last assessed as up-to-date: 9 June 2009.

Date Event Description

24 April 2009 New search has been performed Literature search re-run; no new studies were identiﬁed.

Gastro-oesophageal reflux treatment for prolonged non-specific cough in children and adults (Review) 51 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. HISTORY Protocol first published: Issue 3, 2004 Review first published: Issue 2, 2005

Date Event Description

11 April 2008 New search has been performed Literature search re-run; four new studies were identi- ﬁed which met the eligibility criteria of the review. The conclusions were not altered on the basis of the new trial evidence

5 April 2008 Amended Converted to new review format.

1 August 2007 New citation required but conclusions have not One new study met the entry criteria of the review changed (Steward 2004). This study was identiﬁed from a systematic review on laryngopharyngeal reﬂux and cough (Gatta 2007). It contributed data to a secondary outcome (change in cough scores at end of intervention), but did not alter the conclusions of the review

6 September 2006 New citation required but conclusions have not One new study met the entry criteria of the review changed (Vaezi 2006). This study contributed data to the primary outcome (still coughing at end of trial), but did not alter the conclusions of the review

CONTRIBUTIONSOFAUTHORS A Chang: initiation, design and direction of review, writing of protocol and review, selection of studies, data extraction, performed analysis and interpretation of results. LG: selection of studies and data extraction, writing of primary protocol and review. JG: data extraction; FC: review of manuscript. TJL: data extraction and analysis, interpretation of results and writing of review. The on-line editor for this review was Dr. M Mckean.

DECLARATIONSOFINTEREST Nil.

Gastro-oesophageal reﬂux treatment for prolonged non-speciﬁc cough in children and adults (Review) 52 Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. SOURCES OF SUPPORT Internal sources • Royal Children’s Hospital Foundation, Brisbane, Australia.

External sources • National Health and Medical Research Council, Australia. • Cochrane Child Health Field Bursary Scheme, Canada. • Queensland Government Smart State Funds, Australia. Salary Support for AC

INDEX TERMS

Medical Subject Headings (MeSH) Chronic Disease; Cough [complications; ∗therapy]; Gastroesophageal Reﬂux [complications; ∗therapy]; Randomized Controlled Trials as Topic

MeSH check words Adult; Child; Humans